Psychological Assessment

The Child PTSD Symptom Scale: An Update and Replication of Its Psychometric Properties Reginald D. V. Nixon, Richard Meiser-Stedman, Tim Dalgleish, William Yule, David M. Clark, Sean Perrin, and Patrick Smith Online First Publication, July 1, 2013. doi: 10.1037/a0033324

CITATION Nixon, R. D. V., Meiser-Stedman, R., Dalgleish, T., Yule, W., Clark, D. M., Perrin, S., & Smith, P. (2013, July 1). The Child PTSD Symptom Scale: An Update and Replication of Its Psychometric Properties. Psychological Assessment. Advance online publication. doi: 10.1037/a0033324 Psychological Assessment © 2013 American Psychological Association 2013, Vol. 25, No. 3, 000 1040-3590/13/$12.00 DOI: 10.1037/a0033324

BRIEF REPORT

The Child PTSD Symptom Scale: An Update and Replication of Its Psychometric Properties

Reginald D. V. Nixon Richard Meiser-Stedman and Tim Dalgleish Flinders University Medical Research Council Cognition and Brain Sciences Unit, Cambridge, United Kingdom

William Yule David M. Clark King’s College London University of Oxford

Sean Perrin Patrick Smith Lund University and King’s College London King’s College London

The psychometric properties of the Child PTSD Symptom Scale (CPSS) were examined in 2 samples. Sample 1 (N ϭ 185, ages 6–17 years) consisted of children recruited from hospitals after accidental injury, assault, and road traffic trauma, and assessed 6 months posttrauma. Sample 2 (N ϭ 68, ages 6–17 years) comprised treatment-seeking children who had experienced diverse traumas. In both samples psychometric properties were generally good to very good (internal reliability for total CPSS scores ϭ .83 and .90, respectively). The point-biserial correlation of the CPSS with posttraumatic disorder (PTSD) diagnosis derived from structured clinical interview was .51, and children diagnosed with PTSD reported significantly higher symptoms than non-PTSD children. The CPSS demonstrated applicability to be used as a diagnostic measure, demonstrating sensitivity of 84% and specificity of 72%. The performance of the CPSS Symptom Severity Scale to accurately identify PTSD at varying cutoffs is reported in both samples, with a score of 16 or above suggested as a revised cutoff.

Keywords: CPSS, posttraumatic stress disorder, test–retest reliability, reliable change index, confirma- tory factor analysis

Supplemental materials: http://dx.doi.org/10.1037/a0033324.supp

The Child PTSD Symptom Scale (CPSS; Foa, Johnson, Feeny, quency of the 17 PTSD symptoms (DSM–IV) on a 4-point Likert- & Treadwell, 2001) was developed as a self-report measure of type scale with responses of 0 (not at all),1(once a week or less), posttraumatic stress disorder (PTSD) severity that aligned with the 2(2 to 4 times a week)and3(5 or more times per week). The diagnostic criteria outlined in the Diagnostic and Statistical Man- impact of symptoms on daily function is also measured with seven ual of Mental Disorders (4th ed.; DSM–IV; American Psychiatric dichotomously scored items that tap areas such as schoolwork and Association, 1994). The CPSS was designed for use with children relationship with one’s family. The original format has good This document is copyrighted by the American Psychological Association or one of its allied8–18 publishers. years of age, is relatively quick to administer and generally psychometric properties (e.g., internal consistency of .89 for the This article is intended solely for the personal use ofseems the individual user and is not to be disseminated broadly. well understood by children. The CPSS indexes the fre- total score) and test–retest reliability over a 1- to 2-week interval

This research was supported by grants from the Australian Rotary Reginald D.V. Nixon, School of , Flinders University, Ad- Health Research Fund and Channel 7 Research Foundation awarded elaide, South Australia, Australia; Richard Meiser-Stedman and Tim Dal- to Reginald D. V. Nixon. Richard Meiser-Stedman was supported by gleish, Medical Research Council Cognition & Brain Sciences Unit, Cam- an MRC Research Studentship (G78/6730) and a Peggy Pollak Re- bridge, United Kingdom; William Yule, Institute of Psychiatry, King’s search Fellowship in Developmental Psychiatry awarded by the Psy- College London, London, United Kingdom; David M. Clark, Department chiatry Research Trust. The authors report no declarations of financial of , University of Oxford, Oxford, United King- interest. dom; Sean Perrin, Institutionen för Psykologi, Lund University, Lund, Correspondence concerning this article should be addressed to Reginald Sweden, and Institute of Psychiatry, King’s College London; Patrick D. V. Nixon, School of Psychology, Flinders University, P.O. Box 2100, Smith, Institute of Psychiatry, King’s College London. Adelaide, SA 5001, Australia. E-mail: [email protected]

1 2 NIXON ET AL.

is acceptable (r ranging between .63 and .85 for the total score and trauma. Approximately half this sample came from Australia and subscales). In terms of validity, the original CPSS scores corre- the other half the United Kingdom, with data from the present lated highly with an established measure of child PTSD symptoms, report representing their 6-month posttrauma assessment (Meiser- the Child Posttraumatic Stress Reaction Index (CPTSD–RI; Py- Stedman, Smith, Glucksman, Yule, & Dalgleish, 2008; Meiser- noos et al., 1987), r ϭ .80 (Foa et al., 2001). The CPSS can be used Stedman, Yule, Smith, Glucksman, & Dalgleish, 2005; Nixon, as a continuous measure of symptom severity (summation of Items Ellis, Nehmy, & Ball, 2010). Children and adolescents who pre- 1–17 with possible scores ranging from 0 to 51), and in a sample sented for treatment at the researchers’ traumatic stress clinics and of earthquake victims (N ϭ 75) assessed 2 years posttrauma, a were diagnosed with PTSD (N ϭ 68, age 6–17 years) made up cutoff of 11 or greater was found to have sensitivity of 95% and Sample 2, and these data were also derived evenly from Australia specificity of 96%. Foa et al. (2001) also reported that the 17 and the United Kingdom (Nixon et al., 2012; Smith et al., 2007). symptom items could be scored dichotomously to generate a The published articles accompanying these data sets detail in full DSM–IV consistent diagnosis of PTSD. inclusion and exclusion criteria; briefly, children had to have been While the CPSS has been used in a number of studies since its exposed to an event sufficient to cause PTSD, be able to complete development, including child PTSD treatment research (e.g., assessments in English, and not be suspected of experiencing Nixon, Sterk, & Pearce, 2012; Smith et al., 2007), surprisingly ongoing trauma (e.g., abuse). Key exclusion criteria included the there has been only modest evaluation of its psychometric prop- presence of organic brain damage or learning difficulties, signifi- erties. In particular, whether the original recommended cutoff cant loss of consciousness at the time of the trauma, or unstable score of 11 to diagnose probable PTSD holds in other trauma types medication regimen (especially in the case of the treatment sample, and samples is unknown. The CPSS was originally compared with Sample 2). The pooling of data from several studies has been another self-report measure (CPTSD–RI), and while one study has effectively used in the child traumatic stress field (see, e.g., Dal- reported its correlation with clinical diagnosis (Rachamim, Help- gleish et al., 2008; Kassam-Adams et al., 2012; Meiser-Stedman et man, Foa, Aderka, & Gilboa-Schechtman, 2011), the CPSS has yet al., 2009) with such integrative data analysis approaches affording to be validated against a structured clinical PTSD interview in advantages of increased statistical power and sample heterogeneity terms of sensitivity and specificity metrics. Such a comparison not (Curran & Hussong, 2009; Kassam-Adams et al., 2012). Table 1 only has implications for the measure’s validity but could have shows the demographic and trauma-related characteristics of the significant cost-efficiency implications if it can be shown that the sample. CPSS converges with structured interviews that are lengthy to administer and resource intensive (e.g., requiring interviewer train- Measures ing). Rachamim et al. (2011) reported comparable The psychometric properties of the CPSS were detailed earlier with a Hebrew version of the original measure in a sample of in this article. PTSD diagnosis was established in the Australian Israeli treatment-seeking youth (N ϭ 156, ages 8–18), 78% of sample with the Clinician Administered PTSD Scale for Children whom had been diagnosed PTSD. While the CPSS showed mod- (CAPS–CA; Nader et al., 1998) and with the Anxiety Disorders erate correlation with PTSD diagnosis (.54), as measured with the Interview Schedule, Child and Parent Report version (ADIS–C/P; 1 Schedule of Affective Disorders and for School Silverman & Albano, 1996) in the United Kingdom sample. Age Children, Revised for DSM–IV (K–SADS–R; Kaufman et al., Structured clinical interviews are currently considered the gold 1997), the utility of the original cutoff score was not reported. standard for child PTSD assessment. While there is limited pub- Kohrt et al. (2011) examined the cross-cultural validity of the lished information on the psychometric properties of the CAPS– CPSS in Nepalese children exposed to war and displacement (N ϭ CA, its format is a replication of the adult form, which has strong 162). However, substantial modification to the original measure reliability and validity data (see Weathers, 2004, for review). In and validation of the measure against need for treatment rather unpublished work (Ellis, 2008), internal reliability for the ␣ϭ than PTSD diagnosis per se precludes comparison of the psycho- CAPS–CA scores was good ( .88 for total symptom severity). metrics of this version. We found interrater reliability performed on 24 randomly selected Based on the consolidation of several data sets obtained from interviews across Samples 1 and 2 to be excellent (100% diagnos- our published and unpublished work, we were in a position to tic agreement; Nixon et al., 2010, 2012). The ADIS–C/P is an This document is copyrighted by the American Psychological Association or one of its allied publishers. examine the CPSS to answer some of the above questions. Ac- established diagnostic tool for child anxiety disorders, including This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. cordingly, we were able to examine the psychometric properties of PTSD, and its scores have been shown to possess excellent test– the CPSS in a sample of single-incident trauma survivors (Sample retest reliability (Silverman, Saavedra, & Pina, 2001). Interrater 1) and to test the sensitivity of different cutoff scores in PTSD- reliability was similarly excellent (100% diagnostic agreement) for treatment-seeking children (Sample 2). 21 interviews in Sample 1 (as reported in Meiser-Stedman et al., 2005). Smith et al. (2007) observed satisfactory interrater reliabil- Method 1 Consistent with the data pooling approach described earlier, samples Participants were collapsed to increase power of the study, which is especially pertinent when examining sensitivity and specificity psychometrics where base rates Two samples were selected from several available data sets. of psychopathology may be low. Examination of the psychometric prop- ϭ erties of the CPSS when divided by sample country of origin revealed Sample 1 consisted of children and adolescents (N 185, age negligible differences, including results where PTSD diagnosis was de- 6–17 years) recruited from the emergency department or pediatric rived by either the CAPS–CA or ADIS; hence, data from the pooled inpatient ward of metropolitan hospitals following single-incident samples is reported. CHILD PTSD SYMPTOM SCALE 3

Table 1 Demographic and Trauma-Related Characteristics

Sample 1 (N ϭ 185) Sample 2 (N ϭ 68) Variable M (SD) M (SD)

Age 12.35 (2.73) 12.21 (3.18) Male sex 58% (n ϭ 107) 60% (n ϭ 41) Ethnicity Caucasian 61% (n ϭ 112) 79% (n ϭ 54) Other 38% (n ϭ 71) 21% (n ϭ 14) Missing 1% (n ϭ 2) Trauma type Road traffic and transport accidents 53% (n ϭ 99) 41% (n ϭ 28) Assault 22% (n ϭ 40) 25% (n ϭ 17) Accidental injuries (including sport) 22% (n ϭ 40) — Other 3% (n ϭ 6) 34% (n ϭ 23)a CPSS severity Re-experiencing 2.16 (2.83) 7.79 (3.70) Avoidance 3.59 (4.19) 9.85 (4.37) Arousal 3.74 (3.62) 9.81 (3.36) Total 10.06 (9.80) 27.44 (9.33) CPSS impairmentb 1.03 (1.59) 2.94 (1.97) Note. CPSS ϭ Child PTSD Symptom Scale. a “Other” category for Sample 2 included house fires, witnessing assault, dog attacks, etc. b Number of impairment items endorsed. The present data used only six of the seven impairment items, omitting item “Doing your prayers.”

ity for PTSD diagnosis across 30 randomly selected interviews Severity of PTSD Symptoms and PTSD Status (␬ϭ.82; Sample 2). Consistent with a general consensus in the child PTSD literature that current DSM–IV PTSD criteria are Table 1 summarizes children’s self-reported PTSD symptom overly conservative when applied to children, children in the severity on the CPSS. Scores in Sample 1 (e.g., CPSS total score, ϭ ϭ present study were coded as PTSD positive if they met either full M 10.06, SD 9.80) were only marginally higher than that ϭ ϭ DSM–IV criteria or subthreshold criteria (defined as meeting two reported by Foa et al. (2001; M 7.60, SD 8.10). Not surpris- of the possible three symptom clusters of re-experiencing, avoid- ingly, scores in the treatment-seeking sample (Sample 2) were ϭ ϭ ance, and hyperarousal, as well as satisfying impairment criteria).2 considerably higher (CPSS total score, M 27.44, SD 9.33). ϭ All analyses reporting PTSD diagnosis include children with full Using structured clinical interviews, 10% (n 19) of children in or subthreshold PTSD. Sample 1 were diagnosed with PTSD, and there were no signifi- cant differences between genders or ethnicities in PTSD rates. All children in Sample 2 had PTSD. Procedure Age and Gender3 All children and caregivers provided written informed assent/ consent at each research site. Children completed self-report ques- Age differences were observed only on the Re-experiencing tionnaires, with the CPSS indexed to the event that led to their subscale in Sample 1, where younger children (6–11 years old; hospital contact (Sample 1) or for which they were seeking treat- ment (Sample 2). Trained interviewers (clinical or 2 advanced students) administered either the Subthreshold or alternative diagnostic criteria used in the child PTSD literature has included the definition adopted in the present study (e.g., CAPS–CA or ADIS to derive diagnostic status in both samples. Meiser-Stedman et al., 2005): a requirement for at least one symptom each This document is copyrighted by the American Psychological Association or one of its allied publishers. of re-experiencing, avoidance, and hyperarousal (e.g., Kassam-Adams & This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Winston, 2004; Kenardy, Spence, & Macleod, 2006), or one symptom of Results and Discussion re-experiencing, one of avoidance, and two of hyperarousal (Meiser- Stedman et al., 2008; Scheeringa, Wright, Hunt, & Zeanah, 2006).We Data were screened for normality and outliers. As expected for found prevalence rates were comparable regardless of which criteria were Sample 1 (a non-treatment-seeking sample with the majority of used. children demonstrating good adjustment), CPSS scores tended to 3 At the suggestion of one reviewer, we examined the psychometric be positively skewed. Analyses on transformed data did not alter properties of the CPSS in Sample 1 for two age groups: children (6- to 11-year-olds) and adolescents (12- to 17-year-olds). As can be seen in the the pattern of findings, with very little absolute change observed supplemental materials available online, the psychometric properties for on correlations of interest. Similarly, extreme scores only just the two age groups were remarkably similar and consistent with findings reached the conventions used to define outliers (e.g., Tabachnick when the sample was considered as a whole. The only minor impact of this & Fidell, 2001); thus, adjusting these had little effect. In particular, age segregation appeared to be in relation to cutoff scores, where some differences were observed relative to when the entire Sample 1 was modifying outliers had no impact on cutoff findings. Data for examined. However, this is more likely to be due small cell sizes for PTSD Sample 2 were normally distributed without significant outliers. positive cases (where percentages can be artificially skewed) rather than Accordingly, raw data are reported throughout. age effects per se. 4 NIXON ET AL.

M ϭ 2.95, SD ϭ 3.20, n ϭ 82) reported significantly more CPSS frequency scores of 0 versus 1–3 (as done by Foa et al., re-experiencing symptoms than middle-school children (12–14 2001) as well as dichotomizing scores of 0–1 versus 2–3. When years old; M ϭ 1.12, SD ϭ 1.94, n ϭ 55, p Ͻ .001). In the we used the 0 versus 1–3 criterion without requiring an impairment treatment-seeking sample (Sample 2), older children (15–17 years item to be endorsed, the sensitivity of the CPSS was high in old, n ϭ 24) reported significantly higher total severity scores Sample 1, identifying 95% (n ϭ 18) of the children diagnosed with (M ϭ 31.53, SD ϭ 8.01) than younger children did (M ϭ 24.31, PTSD by interview. Specificity, however was low (51%, n ϭ 85). ϭ ϭ ϭ SD 8.96, n 32, p .003). These older children also reported Requiring at least one impairment item to be endorsed improved ϭ higher avoidance and impairment scores (avoidance: M 12.00, overall performance, with sensitivity slightly lower (84%, n ϭ 16) ϭ ϭ ϭ Ͻ ϭ SD 3.86, vs. M 8.45, SD 3.80, p .002; impairment: M but specificity increasing to 72% (n ϭ 118). Adopting a stricter 3.87, SD ϭ 1.73, vs. M ϭ 2.16, SD ϭ 1.93, p Ͻ .001). No gender approach to symptom scoring (0–1 vs. 2–3) did not improve differences were observed on CPSS scores in Sample 1 or Sample 2. performance of the CPSS. For example, this stricter scoring with- out the requirement of an impairment item to be endorsed resulted 4 Internal Reliability in sensitivity of 68% (n ϭ 13) and specificity of 84% (n ϭ 139). Internal reliability (Cronbach’s alpha) was very good in Sample Requiring impairment to be endorsed further reduced sensitivity ϭ ϭ 1 (total score, .90; re-experiencing, .84; avoidance, .78; and (58%, n 11), whereas specificity further improved (88%, n arousal, .79). Intercorrelations between the total CPSS score and 145). subscales were also good (re-experiencing, .73; avoidance, .89; arousal, .88) and subscale intercorrelations ranged from .58 to .69. Discriminant Validity Internal reliability was lower but within acceptable limits in Sam- ple 2 (total score, .83; re-experiencing, .78; avoidance, .65; CPSS severity was examined against clinical diagnosis using arousal, .67). The total CPSS score correlated highly with each receiver–operator characteristic (ROC) curves, with the area under subscale (re-experiencing, .75; avoidance, .87; arousal, .83), and the curve (AUC) significant in Sample 1 (AUC ϭ .89, p Ͻ .001). subscale intercorrelations ranged from .42 to .64. ROC analysis was inappropriate for Sample 2 as all children were PTSD positive. Table 2 details the performance of varying cutoff Test–Retest Reliability scores for both samples, including the original published cutoff of 11 (Foa et al., 2001). Given that Sample 1 contained a significant For a subset of Sample 1 the CPSS had been administered 3 subset of children who presented with accidental injury that varied months and 6 months posttrauma (n ϭ 113). On the one hand, in severity, the performance of the CPSS was also examined in just knowing the test–retest reliability over such an interval is useful those who had suffered assault or road traffic trauma. when one might wish to calculate reliability of change indices Foa et al. (2001) reported achieving high sensitivity (95%) and (Jacobson, Roberts, Berns, & McGlinchey, 1999). For example, specificity (96%) using a cutoff score of 11. As is apparent from this interval matches the length of a treatment intervention, as opposed to the relatively short test–retest interval (i.e., 1–2 weeks) Table 2, such levels were not achieved in the current samples. that is typically reported in psychometric studies. On the other However the data suggest that a higher cutoff score (e.g., 16) could hand, it should be recognized that there can be a great deal of be used while still obtaining an optimal balance of sensitivity and change in symptoms in the first 6 months following trauma in specificity (sensitivity between .84 and .93 across Sample 1 and 2; children (Le Brocque, Hendrikz, & Kenardy, 2010), and the sam- specificity of .83 in Sample 1). It should also be remembered that ple was non-treatment-seeking; thus, one might expect lower sta- Foa et al. compared the CPSS against another self-report measure, bility relative to a short test–retest interval. Consistent with this, whereas the present data included diagnostic status derived from test–retest reliability coefficients were good for the total score structured clinical interviews. (although lower than the .84 reported by Foa et al., 2001) and Severity scores on the CPSS clearly discriminated between modest for subscales (total score: .75; re-experiencing, .50; avoid- those diagnosed with PTSD and those who were non-PTSD. As ance, .62; arousal, .70). summarized in Table 3, children in Sample 1 with PTSD had significantly higher scores on all subscales and the total CPSS This document is copyrighted by the American Psychological Association or one of its allied publishers. score than non-PTSD children. This article is intended solely for the personal use ofConvergent the individual user and is not to be disseminated broadly. Validity A point-biserial correlation was calculated between the CPSS total score and PTSD diagnosis in Sample 1, showing a significant 4 The same reviewer suggested conducting factor analysis for Sample 1. relationship (r ϭ .51, p Ͻ .001), which was comparable to that A comprehensive report on the factor analytic structure of the CPSS has reported by Rachamim et al. (2011; r ϭ .54). The CPSS can be been previously published (see Kassam-Adams, Marsac, & Cirilli, 2010) in scored according to DSM–IV criteria for use as a diagnostic mea- a similar, albeit larger, sample, and Sample 1 was modest for the purposes of factor analysis. However, confirmatory factor analysis was conducted. sure. Accordingly, we examined how a diagnosis derived from the As reported in the supplemental materials (see Table S5), none of the CPSS using the full and subthreshold criteria outlined earlier models tested reached adequate fit criteria. However the pattern of fit compared against PTSD diagnosis obtained via structured clinical indices across tested models were similar to those observed by Kassam- interview. Whether requiring impairment items to be endorsed Adams et al. (2010; e.g., the four-factor models commonly labeled as “numbing” and “dysphoria” in the literature were significantly better than impacted on this was also examined. Similarly, we tested whether the DSM–IV factor model). This suggested the most likely explanation of the stringency of the criteria required for a symptom to be con- less than adequate fit was due to sample size limitations (see supplemental sidered endorsed affected diagnostic rates. That is, we examined materials for further details). CHILD PTSD SYMPTOM SCALE 5

Table 2 Correspondence of CPSS With PTSD Status Across Different Cutoff Scores

Sample 1a Sample 1b Sample 2c,d CPSS cutoff (Ն) Sensitivity Specificity Sensitivity Specificity Sensitivity

10 89% (17/19) 65% (108/166) 88% (15/17) 57% (70/122) 99% (67/68) 11 89% (17/19) 66% (110/166) 88% (15/17) 59% (72/122) 99% (67/68) 12 84% (16/19) 69% (114/166) 94% (16/17) 61% (75/122) 97% (66/68) 13 84% (16/19) 72% (120/166) 94% (16/17) 66% (81/122) 97% (66/68) 14 84% (16/19) 76% (126/166) 94% (16/17) 69% (84/122) 96% (65/68) 15 84% (16/19) 80% (132/166) 94% (16/17) 73% (89/122) 94% (64/68) 16 84% (16/19) 83% (137/166) 94% (16/17) 76% (93/122) 93% (63/68) 17 84% (16/19) 83% (138/166) 94% (16/17) 77% (94/122) 91% (62/68) 18 79% (15/19) 86% (143/166) 94% (16/17) 81% (99/122) 88% (60/68) 19 79% (15/19) 87% (145/166) 94% (16/17) 83% (101/122) 85% (58/68) 20 79% (15/19) 89% (148/166) 94% (16/17) 85% (104/122) 85% (58/68) Note. CPSS ϭ Child PTSD Symptom Scale; PTSD ϭ posttraumatic stress disorder. a Full hospital sample (N ϭ 185; 19 PTSD, 166 non-PTSD). b Hospital sample (road traffic trauma and assault only, N ϭ 139; 17 PTSD, 122 non-PTSD). c Treatment sample (N ϭ 68, all PTSD). d Specificity not applicable for treatment sample, as all participants had PTSD.

Functional Impairment CPSS severity being correlated with diagnostic status and children with PTSD demonstrating significantly higher CPSS scores than Table 1 details the descriptive data for severity of impairment non-PTSD children, the utility of the CPSS as a quasidiagnostic for both samples. Internal reliability for both samples was accept- instrument was also illustrated. This has significant practical im- able (.80 for Sample 1; .75 for Sample 2). Test–retest reliability plications, suggesting that when time or resource issues preclude over a 3-month interval for a subset of participants in Sample 1 formal diagnostic assessment with lengthy clinical interviews, the (n ϭ 113) was acceptable (r ϭ .60, p Ͻ .001). CPSS could be used as a proxy. As expected, in Sample 1 total impairment scores were signif- Most of the psychometric properties observed in the original icantly correlated with total CPSS severity (r ϭ .58, p Ͻ .001), and development report of the CPSS were replicated in the present each of the CPSS subscales (re-experiencing, r ϭ .30, p Ͻ .001; study. Some slight differences were observed. For example, test– avoidance, r ϭ .51, p Ͻ .001; and arousal, r ϭ .57, p Ͻ .001). retest correlations in the present data were lower; however, this is Similar patterns were observed in Sample 2 with impairment likely to reflect the difference in sample and methods. Foa et al. scores correlated with total CPSS severity (r ϭ .58, p Ͻ .001), and (2001) used a 1- to 2-week test interval, 2 years posttrauma. each of the CPSS subscales (re-experiencing, r ϭ .26, p ϭ .04; Children in Sample 1 in our study were assessed over a 3-month avoidance, r ϭ .61, p Ͻ .001; and arousal, r ϭ .54, p Ͻ .001). interval, and at 3- and 6-month posttrauma assessments. There Those diagnosed with PTSD reported significantly more impair- typically is further remission of symptoms in the first 6 months ment than children without PTSD (Table 3). There were no age or gender differences for impairment scores in Sample 1 or Sample 2. following trauma in children, with relatively little change from 6 months to 2 years (Le Brocque et al., 2010), which likely accounts for the reduced stability observed in our data. Other differences General Comments observed in the current report included lack of gender effects but The aim of the present article was to examine some of the some age effects, whereas the reverse was observed by Foa et al. psychometric properties of the CPSS, which was developed over While sensitivity and specificity metrics in the current data were 10 years ago. The present data show that the CPSS continues to good, they were not as high as those obtained by Foa et al. (2001), perform well as a measure of self-reported PTSD symptoms. A especially in Sample 1. As mentioned previously, the present strength of the current data is its comparison against diagnostic report measured discriminant validity against a structured clinical This document is copyrighted by the American Psychological Association or one of its alliedstatus publishers. obtained via structured clinical interview. In addition to interview, arguably a more rigorous assessment than the self-report This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Table 3 Means (and Standard Deviations) for CPSS Scores in Sample 1 by Diagnostic Status

CPSS scale PTSD Non-PTSD F Cohen’s d

1.05 ءRe-experiencing 5.47 (4.38) 1.78 (2.32) 34.35 1.44 ءAvoidance 9.32 (5.21) 2.93 (3.53) 50.05 1.65 ءArousal 8.58 (3.31) 3.18 (3.23) 47.42 1.74 ءTotal 24.73 (10.40) 8.38 (8.23) 63.54 0.96 ءCPSS impairment 2.44 (1.80) 0.86 (1.48) 18.56 Note. PTSD, n ϭ 19; non-PTSD, n ϭ 166. CPSS ϭ Child PTSD Symptom Scale; PTSD ϭ posttraumatic stress disorder. .p Ͻ .001 ء 6 NIXON ET AL.

measure used by Foa et al. Important sample differences were also stress disorder in children and adolescents. British Journal of Psychiatry, apparent. Sample 1 in particular contained fewer Caucasian chil- 192, 392–393. doi:10.1192/bjp.bp.107.040451 dren than the original normative sample (61% vs. 89%) and more Ellis, A. A. (2008). Cognitive and social support factors in ASD, PTSD and boys (58% vs. 41%), and both Sample 1 and 2 included children in children and adolescents following single-incident trauma ages 6–7 as well as 16–17, compared with children in the 8- to (Unpublished doctoral dissertation). Flinders University, Adelaide, Aus- tralia. 15-year-old age range. Foa et al.’s sample also comprised children Foa, E. B., Johnson, K. M., Feeny, N. C., & Treadwell, K. R. (2001). The who had experienced the same event (earthquake) 2 years previ- Child PTSD Symptom Scale: A preliminary examination of its psycho- ously, whereas in the present study mixed types of traumatic metric properties. Journal of Clinical Child Psychology, 30, 376–384. experiences constituted the stressful events, typically 6 months doi:10.1207/S15374424JCCP3003_9 prior to assessment. It is likely that some of these differences Jacobson, N. S., Roberts, L. J., Berns, S. B., & McGlinchey, J. B. (1999). contributed to the lower sensitivity and specificity observed in the Methods for defining and determining the clinical significance of treat- present data. Of course, the clinician’s goal for using the CPSS ment effects: Description, application, and alternatives. Journal of Con- would guide the selection of an appropriate cutoff. If it is used for sulting and Clinical Psychology, 67, 300–307. doi:10.1037/0022-006X screening purposes where there are the resources to manage po- .67.3.300 tential false positives, a score of 16 seems sufficient, and we Kassam-Adams, N., Marsac, M., & Cirilli, C. (2010). Posttraumatic stress suggest that in most cases clinicians would err on the side of disorder symptom structure in injured children: Functional impairment identifying children with probable PTSD. While near perfect sen- and depression symptoms in a confirmatory factor analysis. Journal of the American Academy of Child & Adolescent Psychiatry, 49, 616–625. sitivity (99%) was observed in our treatment seeking sample using doi:10.1097/00004583-201006000-00010 Foa et al.’s cutoff of 11, we lacked specificity data, and it is worth Kassam-Adams, N., Palmieri, P., Rork, K., Delahanty, D., Kenardy, J., noting that sensitivity for a cutoff of 16 was very good (93%). Kohser, K.,...Bui, E. (2012). Acute stress symptoms in children: Taking into account these factors, the fact that we used a diagnos- Results from an international data archive. Journal of the American tic clinical interview for comparison purposes, and our clinical Academy of Child and Adolescent Psychiatry, 51, 812–820. http://dx. experience with using the CPSS in a variety of contexts, we argue doi.org/10.1016/j.jaac.2012.05.013 that a cutoff of 11 is probably too low; thus, we recommend Kassam-Adams, N., & Winston, F. K. (2004). Predicting child PTSD: The considering a score of at least 16 as the cutoff. Considering all the relationship between (ASD) and PTSD in injured above issues, the discrepancy in the other achieved psychometrics children. Journal of the American Academy of Child & Adolescent (internal reliability, intercorrelations between subscales, etc.) be- Psychiatry, 43, 403–411. doi:10.1097/00004583-200404000-00006 tween the current data and the original development study is Kaufman, J., Birmaher, B., Brent, D., Rao, U., Flynn, C., Moreci, P.,... relatively small, underscoring the robustness of the CPSS as a Ryan, N. (1997). Schedule for Affective Disorders and Schizophrenia for School-Age Children—Present and Lifetime Version (K–SADS– measure of PTSD. PL): Initial reliability and validity data. Journal of the American Acad- Several limitations are acknowledged. We did not have consis- emy of Child & Adolescent Psychiatry, 36, 980–988. doi:10.1097/ tent measures across data sets for other psychopathology (depres- 00004583-199707000-00021 sion, general anxiety); thus, we could not test the divergent validity Kenardy, J. A., Spence, S. H., & Macleod, A. C. (2006). Screening for of the CPSS. Prior studies of the CPSS (e.g., Foa et al., 2001; posttraumatic stress disorder in children after accidental injury. Pediat- Rachamim et al., 2011) have found good support for its divergent rics, 118, 1002–1009. doi:doi.org/10.1542/peds.2006-0406 validity, and prior publication of the current data sets (e.g., Meiser- Kohrt, B. A., Jordans, M. J. D., Tol, W. A., Luitel, N. P., Maharjan, S. M., Stedman et al., 2005, 2008; Nixon et al., 2010, 2012; Smith et al., & Upadhaya, N. (2011). Validation of cross-cultural child mental health 2007) has documented levels of anxiety and depression in the and psychosocial research instruments: Adapting the Depression Self- samples. Data were collected prior to proposed DSM-5 criteria for Rating Scale and Child PTSD Symptom Scale in Nepal. BMC Psychi- PTSD; thus, we were not in a position to examine how these atry, 11, 127. doi:doi.org/10.1186/1471-244X-11-127 additional symptoms might impact on performance of the CPSS. Le Brocque, R. M., Hendrikz, J., & Kenardy, J. A. (2010). The course of posttraumatic stress in children: Examination of recovery trajectories Prior research has examined child–caregiver agreement on symp- following traumatic injury. Journal of Pediatric Psychology, 35, 637– toms and diagnosis, especially in younger children (Meiser- 645. doi:10.1093/jpepsy/jsp050 Stedman et al., 2008). Given that the CPSS was completed by Meiser-Stedman, R., Smith, P., Bryant, R., Salmon, K., Yule, W., Dal- children, it remains to be seen whether optimal identification of gleish, T., & Nixon, R. D. V. (2009). Development and validation of the This document is copyrighted by the American Psychological Association or one of its allied publishers. child PTSD would be assisted by incorporating parent or caregiver Child Post-Traumatic Cognitions Inventory (CPTCI). Journal of Child This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. responses. Despite these limitations, the current study indicates Psychology and Psychiatry, 50, 432–440. doi:10.1111/j.1469-7610 that the CPSS is a useful measure of child PTSD. It is short, easy .2008.01995.x for children to understand, easily scored, and can be used as a Meiser-Stedman, R., Smith, P., Glucksman, E., Yule, W., & Dalgleish, T. reasonable proxy for clinical diagnosis. (2008). The posttraumatic stress disorder diagnosis in preschool- and elementary school-age children exposed to motor vehicle accidents. References American Journal of Psychiatry, 165, 1326–1337. doi:10.1176/appi.ajp .2008.07081282 American Psychiatric Association. (1994). Diagnostic and statistical man- Meiser-Stedman, R., Yule, W., Smith, P., Glucksman, E., & Dalgleish, T. ual of mental disorders (4th ed.). Washington, DC: Author. (2005). Acute stress disorder and posttraumatic stress disorder in chil- Curran, P. J., & Hussong, A. M. (2009). Integrative data analysis: The dren and adolescents involved in assaults or motor vehicle accidents. simultaneous analysis of multiple data sets. Psychological Methods, 14, American Journal of Psychiatry, 162, 1381–1383. doi:10.1176/appi.ajp 81–100. doi:10.1037/a0015914 .162.7.1381 Dalgleish, T., Meiser-Stedman, R., Kassam-Adams, N., Ehlers, A., Win- Nader, K. O., Newman, E., Weathers, F. W., Kaloupek, D. G., Kriegler, ston, F., Smith, P.,...Yule, W. (2008). Predictive validity of acute J. A., Blake, D. D., & Pynoos, R. S. (1998). Clinician-Administered CHILD PTSD SYMPTOM SCALE 7

PTSD Scale for Children and Adolescents (CAPS–CA). Boston, MA: Silverman, W. K., & Albano, A. M. (1996). Interview National Center for Posttraumatic Stress Disorder. Schedule for DSM–IV: Child and Parent Interview Schedule. San An- Nixon, R. D. V., Ellis, A. E., Nehmy, T. J., & Ball, S. (2010). Screening tonio, TX: Psychological Corporation. and predicting posttraumatic stress and depression in children following Silverman, W. K., Saavedra, L. M., & Pina, A. A. (2001). Test–retest single-incident trauma. Journal of Clinical Child and Adolescent Psy- reliability of anxiety symptoms and diagnoses with the Anxiety Disor- chology, 39, 588–596. doi:10.1080/15374416.2010.486322 ders Interview Schedule for DSM–IV: Child and parent versions. Jour- Nixon, R. D. V., Sterk, J., & Pearce, A. (2012). A randomized trial of nal of the American Academy of Child & Adolescent Psychiatry, 40, cognitive behaviour therapy and cognitive therapy for children with 937–944. doi:10.1097/00004583-200108000-00016 Journal posttraumatic stress disorder following single-incident trauma. Smith, P., Yule, W., Perrin, S., Tranah, T., Dalgleish, T., & Clark, D. M. of Abnormal Child Psychology, 40, 327–337. doi:10.1007/s10802-011- (2007). Cognitive–behavioral therapy for PTSD in children and adoles- 9566-7 cents: A preliminary randomized controlled trial. Journal of the Amer- Pynoos, R. S., Frederick, C., Nader, K., Arroyo, W., Steinberg, A., Eth, S., ican Academy of Child & Adolescent Psychiatry, 46, 1051–1061. doi: . . . Fairbanks, L. (1987). Life threat and posttraumatic stress in school- aged children. Archives of General Psychiatry, 44, 1057–1063. doi: 10.1097/CHI.0b013e318067e288 10.1001/archpsyc.1987.01800240031005 Tabachnick, B. G., & Fidell, L. S. (2001). Using multivariate statistics (4th Rachamim, L., Helpman, L., Foa, E. B., Aderka, I. M., & Gilboa- ed.). Boston, MA: Allyn & Bacon. Schechtman, E. (2011). Validation of the Child Posttraumatic Symptom Weathers, F. W. (2004). Clinician-Administered PTSD Scale (CAPS): Scale in a sample of treatment-seeking Israeli youth. Journal of Trau- Technical manual. Los Angeles, CA: Western Psychological Services. matic Stress, 24, 356–360. doi:10.1002/jts.20639 Scheeringa, M. S., Wright, M. J., Hunt, J. P., & Zeanah, C. H. (2006). Factors affecting the diagnosis and prediction of PTSD symptomatology Received October 11, 2012 in children and adolescents. American Journal of Psychiatry, 163, 644– Revision received March 4, 2013 651. doi:10.1176/appi.ajp.163.4.644 Accepted April 25, 2013 Ⅲ This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.