The Diagnosis of Posttraumatic Stress Disorder in School-Aged Children and Adolescents Following Pediatric Intensive Care Unit Admission

Abstract

Objectives:

This study explored the diagnosis of posttraumatic stress disorder (PTSD) in children and adolescents following pediatric intensive care unit (PICU) admission. Specifically, the study aimed to describe the presentation and prevalence of PTSD symptoms 6 months postdischarge, explore the validity of the DSM-IV PTSD algorithm and alternative PTSD algorithm (PTSD-AA) in school-aged children and adolescents, and examine the diagnostic utility of Criterion C3 (inability to recall aspects of a trauma) in this cohort.

Methods:

Participants were 59 children aged 6–16, admitted to PICU for at least 8 hours. PTSD was assessed via diagnostic interview (Children's PTSD Inventory) 6 months following PICU discharge.

Results:

The PTSD-AA was found to provide the most valid measure of PTSD at 6 months. Removing Criterion C3 improved the validity of Criterion C.

Conclusions:

This study supports the use of the PTSD-AA excluding Criterion C3 for identifying highly traumatized children and adolescents following PICU admission.

C oncerns have been raised regarding the developmental sensitivity of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV-TR; American Psychiatric Association 2000) posttraumatic stress disorder (PTSD) diagnostic criteria in identifying children suffering significant emotional distress and impairment in daily functioning following trauma exposure (Carrion et al. 2002; Iselin et al. 2010; Meiser-Stedman et al. 2008; Scheeringa et al. 2011). Specifically, the current diagnostic thresholds of one re-experiencing symptom, two hyper-arousal symptoms, and three avoidance symptoms may underestimate the number of children experiencing significant and debilitating posttraumatic responses (Cohen and Scheeringa 2009). It has been argued that children's expression of trauma symptoms may be different to adults due to developing language and abstract cognitive capacities (especially in young children; Scheeringa et al. 2011). Thus the DSM-IV criteria may be too stringent for children given the reliance on verbal descriptions of internalized experiences (Scheeringa et al. 2006; Scheeringa et al. 2011). This is of current significance as the newly released DSM-5 (American Psychiatric Association, 2013) now features a PTSD preschool subtype, recognizing that developmental adjustments should be made to diagnostic criteria for young children aged under 6. However, school-aged children and adolescents remain subject to the new DSM-5 PTSD diagnostic criteria for adults.

Variations to DSM-IV PTSD diagnostic thresholds have been proposed in an attempt to improve the sensitivity of diagnostic algorithms in capturing the symptom manifestation in children. In particular, an alternative algorithm initially proposed for preschool children (PTSD-AA; Scheeringa et al. 1995) modifies DSM-IV PTSD symptom wording to be more behaviorally and developmentally sensitive for young children. It also omits Criterion A2 (subjective distress) due to difficulties in determining the subjective experience of children with limited cognitive and language skills, and lowers the Criterion C (avoidance) threshold from three required symptoms to one (Scheeringa et al. 2003). This modification is consistent with evidence that Criterion C is the cluster least frequently endorsed by children, poorest at loading onto factor analyses, and best at uniquely predicting DSM-IV diagnosis, indicating that the DSM-IV Criterion C threshold is too stringent for children (Scheeringa et al. 2011).

Several studies have now explored PTSD diagnosis and symptomatology in young children using developmentally sensitive criteria (De Young et al. 2011; Levendosky et al. 2002; Meiser-Stedman et al. 2008; Ohmi et al. 2002; Scheeringa et al. 2006; Scheeringa and Zeanah 2008; Scheeringa et al. 1995; Scheeringa et al. 2003; Stoddard et al. 2006). These studies provide compelling evidence that PTSD-AA shows better diagnostic validity and sensitivity than DSM-IV PTSD criteria in this age group. Far fewer studies have examined the validity of alternative algorithms in school-aged children and adolescents. However, preliminary investigations suggest that the PTSD-AA may have better validity for older youth as well.

Meiser-Stedman and colleagues (2008) compared PTSD-AA and DSM-IV diagnoses in 7–10-year-old children who had been admitted to the hospital following a motor vehicle accident. They reported that the frequency of PTSD-AA was higher than DSM-IV for older children, consistent with patterns found in younger children. Children identified as PTSD positive using the PTSD-AA displayed similar symptoms frequencies as those identified using the DSM-IV algorithm, suggesting that children meeting PTSD-AA criteria were similarly symptomatic or distressed. Scheeringa and colleagues (2006) reported that rates of endorsement of Criterion C were substantially different across age groups in injured children, and concluded that the PTSD-AA Criterion C threshold of one required symptom was more appropriate for young children (0–6 years), probably more appropriate for school-aged children (7–11 years), and probably too lenient for adolescents (12–18 years) in comparison to the DSM-IV threshold of three required symptoms. Finally, Iselin and colleagues (2010) reported that the PTSD-AA was superior to the DSM-IV PTSD algorithm in identifying children and adolescents (6–15 years) with impairrment in psychosocial functioning following traumatic brain injury. These three studies suggest that the PTSD-AA may show greater diagnostic validity than the DSM-IV PTSD algorithm for school-aged children and adolescents, although the validity of reducing the Criterion C threshold from three required symptoms to one symptom remains unclear for older age groups.

DSM-IV PTSD Criterion C also presents challenges for accurate diagnosis of PTSD in particular cohorts of traumatized children. Criterion C3, “Inability to recall an important aspect of the trauma,” may be problematic as a diagnostic criterion in children with alterations to consciousness initiated by traumatic brain injury or other medical trauma. This criterion is meant to assess avoidance expressed as psychological dissociation, but lack of detailed memory for aspects of a medical trauma may instead relate to organic, medical, or treatment effects in pediatric patients. Thus, inclusion of this item may result in overdiagnosis in some pediatric populations and criteria that lack specificity in identifying children with high levels of symptomatology and functional impairment. For example, in Iselin et al.'s (2010) study, excluding Criterion C3 from the PTSD-AA improved diagnostic validity in children who had sustained a traumatic brain injury and best identified children suffering psychosocial impairment; PTSD-AA including Criterion C3, and DSM-IV PTSD, with or without Criterion C3, failed to identify children with functional impairment.

No studies have extended this research to other pediatric patients, yet children with a variety of medical conditions also experience alterations in consciousness by virtue of their illness/injury or medical treatment regimens (e.g., delirium, sedation, anesthesia, alterations to usual sleep patterns) (Dow et al. 2012). This criterion is retained in the new DSM-5 (Criterion D1), yet further investigation is required to determine whether this criterion is diagnostically useful in pediatric patients with medical events that may result in alterations in consciousness.

The purpose of this study was to explore the diagnosis of PTSD in school-aged children and adolescents following an admission to the pediatric intensive care unit (PICU). Specifically, we sought to (1) describe the prevalence and presentation of posttraumatic stress in children and adolescents following PICU admission, (2) explore the validity of the PTSD-AA in school-aged children and adolescents, and (3) investigate the diagnostic utility of Criterion C3 in this cohort. It was hypothesized that PTSD-AA would result in a greater frequency of PTSD “cases” than the DSM-IV PTSD algorithm but that school-aged children and adolescents identified by PTSD-AA would be highly symptomatic and functionally impaired. It was also hypothesized that removing Criterion C3 from the PTSD-AA would improve the identification of children with significant symptomatology and impairment.

Method

Participants

This study was approved and conducted in accordance with the University of Queensland Human Ethics Committee and the Royal Children's Hospital Human Research and Ethics Committee. As part of a prospective longitudinal research project investigating the psychological impact of PICU admission on families, surviving children aged 6–16 years admitted to the Royal Children's Hospital PICU, Brisbane, Australia for at least 8 hours (equivalent to an overnight stay) between June 2008 and January 2011 were recruited consecutively. Exclusion criteria were (1) prior PICU admission, (2) length of stay >28 days, (3) posttraumatic amnesia >28 days; (4) non-accidental injury, and (5) developmental delay or intellectual impairment.

Of 196 eligible families, 19 were missed at recruitment (unable to contact), 34 refused (15 – too busy/overburdened, 7 – child refused, 8 – not interested/not relevant, 2 – too distressing, 2 – involved in other research), 37 consented but did not provide data at any assessment (non-consenters), 44 consented but dropped out before the 6 month assessment (6 – dropped out before initial assessment, 21 – did not provide data after initial interview, 5 – unable to contact, 5 – too busy/overburdened, 5 – no concerns, 2 – child died) and 3 had missing interview data at the 6 month assessment. Thus 59 children completed interviews at the 6 month assessment. See Table 1 for sample characteristics.

Table 1.

Sample Characteristics (n=59)

	M (SD)	n (%)
Age, years	10.76 (2.59)
Gender, male		33 (56)
Family of origin, both biological parents^a		40 (78)
Participating parent's highest level of education^a
Did not complete high school		6 (12)
Completed high school		8 (16)
College certificate		18 (35)
University degree		19 (37)
Length of stay in PICU, hrs	45.08 (55.12)
8–24hrs		33 (56)
25–72hrs		15 (25)
≥73hrs		11 (19)
PIM2 Risk of death	1.71 (2.49)
Intubated and ventilated during admission		17 (29)
Reason for admission
Postoperative care		25 (42)
Injury		13 (22)
Respiratory		6 (10)
Other		15 (25)
Elective admission		24 (41)
Any TBI (primary or nonprimary diagnosis)		11 (19)
Sedated during admission		16 (27)
Medications
Benzodiazepines, any		15 (26)
Opiates, any		37 (63)
Hypnotics, any		9 (15)

Data are missing for 8 children. PIM2=Pediatric Index of Mortality, Revised. TBI=traumatic brain injury.

Measures

Child PTSD

The Children's PTSD Inventory (CPTSDI; Saigh et al. 2000) is a DSM-IV-based clinician-administered diagnostic interview assessing PTSD in youth aged 7–18 years. The CPTSDI has high content validity, as established by high ratings of correspondence between CPTSDI items and DSM-IV PTSD diagnostic criteria (Saigh et al. 2000). The CPTSDI also demonstrates excellent inter-rater reliability (α=.95), overall diagnostic agreement (κ=.93–.95), and concurrent validity with other PTSD interviews (Saigh et al. 2000; Yasik et al. 2001). The CPTSDI was conducted by a postgraduate psychologist with experience in assessing PTSD in children. Two graduate-level psychologists independently scored six recorded interviews and 100% diagnostic agreement was obtained.

The CPTSDI also provided a measure of functional impairment in this study. In addition to assessing significant emotional distress, the CPTSDI assesses four functional domains to determine whether a child meets DSM-IV PTSD Criterion E (significant distress/functional impairment):

• “Have you been having more problems with your classmates or other children since your bad experience occurred?” (Item E2),

• “Have your grades in school gotten worse since this happened?” (Item E3),

• “Have you been having more problems with your parents and/or the people that you live with since this happened?” (Item E4),

• “Have you been having more problems with your teachers since this happened?” (Item E4).

A continuous measure of functional impairment (number of impaired domains) was calculated by summing the four functional impairment items. A dichotomous variable was also calculated by endorsement of any of the four domains.

Illness severity

Illness severity was measured by the Revised Paediatric Index of Mortality (PIM2; Slater et al. 2003). The PIM2 is an index that uses a regression model to predict a child's mortality from ten variables collected immediately upon PICU admission. The model has been shown to discriminate well between survivors and children who died in PICU, AUC=0.90 (95% CI=0.89–0.91; Slater, et al., 2003). A PIM2 risk of death of 1.5 indicates a 1.5% chance of mortality.

Procedure

Families were invited to participate in the study face-to-face upon PICU discharge (71/177; 40%) or by letter and follow-up phone call following discharge (106/177; 60%). Written informed consent was obtained from parents and assent from children. Six months postdischarge, children completed the CPTSDI (median=6.9 months, range=5.4–8.7 months). The Royal Children's Hospital PICU serves as a specialist referral base for critical pediatric care across Queensland and Northern New South Wales. Thus, while some interviews were conducted face-to-face at the patient's home (27%) or during outpatient visits (7%), most were conducted over the phone (66%). Demographic data were obtained from parents at an earlier assessment and medical data were obtained from medical records.

Data analysis

All statistical analyses were performed using the Statistical Package for Social Sciences (SPSS 19.0; Chicago, IL). The frequency of PTSD symptoms, criterions and diagnoses in children and adolescents were compared using Pearson's χ² tests.

Four PTSD algorithms were calculated: DSM-IV PTSD, DSM-IV PTSD no C3 (as for DSM-IV PTSD, but excluding Criterion C3), PTSD-AA, (as for DSM-IV PTSD, but excluding Criterion A2 and requiring 1 Criterion C symptom instead of 3), and PTSD-AA no C3 (as for PTSD-AA but excluding Criterion C3). The validity of each algorithm was examined by comparing how well they identified children who were highly symptomatic (assessed by total number of PTSD symptoms) and functionally impaired (assessed by number of impaired functional domains). The diagnostic validity of Criterion C3 was further evaluated by comparing how well various Criterion C thresholds (Crit C (1); one symptom required, Crit C (1) no C3; one symptom required excluding C3, Crit C (3); three symptoms required, Crit C (3) no C3; three symptoms required excluding C3) in identifying children who were highly symptomatic and functionally impaired.

The total number of PTSD symptoms was normally distributed, but the number of impaired functional domains was significantly positively skewed. Thus, independent groups' t-tests were performed to compare mean number of PTSD symptoms across PTSD algorithms and Criterion C thresholds, and Mann-Whitney U-tests were performed to compare the number of impaired domains across PTSD algorithms and Criterion C thresholds.

To further examine the utility of PTSD algorithms, cross tabulations were performed to calculate and compare the sensitivity and specificity of each algorithm in discriminating between children who reported any functional impairment (endorsement of impairment in any of the four functional domains) and those who did not. Fischer's exact test was also performed to determine whether each algorithm significantly differentiated between children with and without impairment in each functional domain.

Results

PTSD symptoms

PTSD symptom, criterion, and algorithm frequencies are presented in Table 2. The symptoms most frequently endorsed by PICU patients were inability to recall aspects of the event, hypervigilance, avoidance of thoughts or feelings, physiological reactivity to trauma reminders, and intrusive thoughts or pictures. Comparison of school-aged children (aged 6–11) and adolescents (aged 12–16) revealed very similar symptom presentation between the two age groups. The only significant differences observed were in avoidance of thoughts or feelings, χ²(1)=5.54, p=.02, with more frequent endorsement by school-aged children than adolescents, and difficulty concentrating, χ²(1)=4.96, p=.03, with more frequent endorsement by adolescents than school-aged children. Non-significant trends were noted in diminished interest in activities, χ²(1)=3.64, p=.06 and hypervigilance, χ²(1)=3.54, p=.06, with more frequent endorsement by adolescents.

Table 2.

PTSD Symptom and Algorithm Frequencies 6 Months Post-PICU Admission

Criterion A	PICU all (n=59)	PICU 6–11 yrs (n=34)	PICU 12–16 yrs (n=25)	χ²
A1: Experienced event that threatened death/ physical integrity	59 (100%)	34 (100%)	25 (100%)	—
A1: Witnessed traumatic event occurring to others	15 (25%)	6 (18%)	9 (36%)	2.56
A2: Fear, helplessness, horror	44 (75%)	24 (71%)	20 (80%)	0.67
A2: Disorganized/agitated behavior	9 (15%)	3 (9%)	6 (24%)	2.56
Criterion A met	45 (76%)	23 (68%)	20 (80%)	1.11
Criterion B (Re-experiencing)
B1: Recurrent, distressing memories	20 (34%)	14 (41%)	6 (24%)	1.90
B2: Recurrent, distressing dreams	13 (22%)	7 (21%)	6 (24%)	0.10
B3: Sense of reliving event	2 (3%)	2 (6%)	0 (0%)	1.52
B4: Psychological distress at reminders	17 (29%)	9 (27%)	8 (32%)	0.22
B5: Physiological reactivity at reminders	20 (34%)	11 (32%)	9 (36%)	0.09
Criterion B met (≥1 symptom)	33 (56%)	20 (59%)	13 (52%)	0.27
Criterion C (Avoidance)
C1: Avoidance of thoughts, feelings	22 (37%)	17 (50%)	5 (20%)	5.54^**
C2: Avoidance of activities, places, people	13 (22%)	7 (21%)	6 (24%)	0.10
C3: Inability to recall aspects of event	50 (85%)	29 (85%)	21 (84%)	0.20
C4: Diminished interest in activities	12 (20%)	4 (12%)	8 (32%)	3.64^*
C5: Detachment/estrangement from others	16 (27%)	9 (27%)	7 (28%)	0.20
C6: Restricted affect	9 (15%)	5 (15%)	4 (16%)	0.20
C7: Sense of foreshortened future	9 (15%)	5 (15%)	4 (16%)	0.20
Criterion C met (≥3 symptoms)	24 (41%)	14 (41%)	10 (40%)	0.01
Criterion C met (≥1 symptom)	53 (90%)	31 (91%)	22 (88%)	0.16
Criterion C met exc C3 (≥3 symptoms)	13 (22%)	8 (24%)	5 (20%)
Criterion C met exc C3 (≥1 symptom)	31 (59%)	20 (59%)	15 (60%)
Criterion D (Hyperarousal)
D1: Difficulty sleeping	14 (24%)	6 (18%)	8 (32%)	1.64
D2: Irritability/anger	16 (27%)	8 (24%)	8 (32%)	0.52
D3: Difficulty concentrating	19 (32%)	7 (21%)	12 (48%)	4.96^**
D4: Hypervigiliance	27 (46%)	12 (35%)	15 (60%)	3.54^*
D5: Exaggerated startle response	8 (14%)	4 (12%)	4 (16%)	0.22
Criterion D met (≥2 symptoms)	26 (44%)	13 (38%)	13 (52%)	1.11
Criterion F (Significant distress/impairment)	25 (42%)	13 (38%)	12 (48%)	0.56
DSM-IV PTSD	15 (25%)	10 (29%)	5 (20%)	0.67
DSM-IV PTSD no C3	10 (17%)	7 (21%)	3 (12%)	0.76
PTSD-AA/PTSD-AA no C3	17 (29%)	10 (29%)	7 (28%)	0.01

p<.10, ^** p<.05.

PTSD-AA=PTSD Alternative Algorithm.

Note: Symptoms are endorsed only if children report they have been persisting beyond four weeks, thus all children who endorsed symptoms met Criterion E.

Criterion C and item C3

As shown in Table 3, all Criterion C thresholds discriminated between high and low symptomatology in children. When three Criterion C symptoms were required, the endorsement rate was 41%. Children meeting this threshold had greater functional impairment than those not meeting the threshold. When three symptoms were required but Criterion C3 was excluded, the criterion also identified children with greater functional impairment. However, it appeared too restrictive as the endorsement rate was low (23%).

Table 3.

Comparison of Symptom Clusters and PTSD Algorithms in Identifying Children with High and Low Symptomatology and Functional Impairment

		No. Impaired Domains, M (SD)				No. PTSD Symptoms, M (SD)
Symptom cluster/ PTSD algorithm	n Cases (total=59)	Case	Non-case	z	p	Case	Non-case	t	p
Crit C (1)	53 (90%)	0.55 (0.97)	0.17 (0.41)	0.93	.454	5.32 (3.53)	0.83 (1.17)	3.08	.003
Crit C (1, no C3)	35 (59%)	0.80 (1.11)	0.08 (0.28)	3.28	.001	7.23 (2.77)	1.42 (0.97)	9.86	<.001
Crit C (3)	24 (41%)	1.00 (1.22)	0.17 (0.45)	4.27	<.001	5.32 (3.53)	0.83 (1.17)	9.70	<.001
Crit C (3, no C3)	13 (22%)	1.00 (0.91)	0.37 (0.90)	3.60	<.001	9.46 (2.11)	3.57 (2.82)	7.00	<.001
DSM-IV PTSD	15 (25%)	1.40 (1.30)	0.20 (0.51)	4.58	<.001	9.53 (1.73)	3.27 (2.56)	6.26	<.001
DSM-IV PTSD no C3	10 (17%)	1.00 (0.94)	0.41 (0.91)	2.66	.008	10.30 (1.57)	3.76 (2.83)	7.06	<.001
PTSD-AA/PTSD-AA no C3	17 (29%)	1.41 (1.23)	0.14 (0.42)	5.26	<.001	9.35 (1.73)	3.05 (2.34)	9.87	<.001

Crit C (1)=Criterion C threshold of 1 required symptom; Crit C (1, no C3)=Criterion C threshold of 1 required symptom excluding Criterion C3; Crit C (3)=Criterion C threshold of 3 required symptoms; Crit C (3, no C3)=Criterion C threshold of 3 required symptoms excluding Criterion C3; DSM-IV PTSD no C3=DSM-IV PTSD diagnosis excluding Criterion C3; PTSD-AA=PTSD Alternative Algorithm; PTSD-AA no C3=PTSD Alternative Algorithm excluding Criterion C.

When only one symptom was required, the criterion appeared too inclusive as 90% of children met the threshold. Furthermore, it failed to identify children with greater functional impairment. However, when one symptom was required but Criterion C3 was excluded, the endorsement rate was 59% and children with greater functional impairment were identified.

PTSD diagnostic algorithms

As shown in Table 3, six months post-PICU, 15 children (25%) met full DSM-IV PTSD diagnostic criteria and 17 (29%) met PTSD-AA criteria. Ten children (17%) met DSM-IV PTSD diagnostic criteria when Item C3 was removed. There were no differences in the number of children diagnosed with PTSD-AA and PTSD-AA no C3.

For all PTSD algorithms, children who met the diagnostic criteria displayed significantly higher symptomatology and greater functional impairment than those who did not fulfill criteria. The two children who met the PTSD-AA but not the DSM-IV algorithm endorsed an average of eight PTSD symptoms and displayed functional impairment in either one or two domains.

Further investigation of the ability of each algorithm to identify children with impairment in any of the four functional domains revealed a sensitivity of .74 for PTSD-AA, .63 for DSM-IV PTSD, and .37 for DSM-IV PTSD no C3 (specificity for all=.93). Fisher's exact test revealed that DSM-IV PTSD and PTSD-AA significantly identified children suffering impairment in each functional domain of decline in school performance (p<.001 for both), problems with classmates (p=.032 and p=.006 respectively), parents (p=.032 for both) and teachers (p=.003 and p=.005 respectively). DSM-IV PTSD no C3 identified children with a decline in school performance (p=.047), but failed to identify those displaying problems with classmates (p=.734), parents (p=.266), or teachers (p=.130).

Discussion

This study explored the diagnosis of PTSD in school-aged children and adolescents following admission to the PICU and makes several contributions to current literature. First, it examined the frequency of PTSD symptoms in children aged 6 and over and demonstrated that few differences are seen in patterns of symptom presentation between school-aged children and adolescents. This information provides a better understanding of the manifestation of posttraumatic stress during this developmental period and may help to guide prevention and treatment of PTSD in this age group.

Second, this study demonstrated that Criterion C3 was not a useful criterion in this cohort. Most children (85%) endorsed this item, suggesting that many children had alterations in consciousness that affected their ability to recall the PICU event. This may be due to their reason for admission (e.g., traumatic brain injury (TBI)), medical complications (e.g., delirium from infection or medication), or effects of therapeutic medications (e.g., sedatives). This finding is likely to apply to other child and adult cohorts with probable alterations in consciousness (e.g., TBI cohorts, heavily medicated/sedated, drugged, alcohol abuse at the time of the trauma). It should be noted that in this cohort, excluding item C3 from the DSM-IV Criterion C threshold of three required symptoms resulted in criteria that were too stringent and failed to identify several children reporting significant symptomatology and impairment. Lowering the Criterion C threshold to one required symptom and excluding C3 resulted in a reasonable endorsement rate, and a Criterion C that discriminated well between children with high and low symptomatogy and functional impairment. This is consistent with our additional findings regarding alternative PTSD criteria for school-aged children and adolescents.

Third, this study provides evidence that an alternative algorithm found to be superior to DSM-IV PTSD in young children is also more diagnostically valid in older children. Few studies to date have examined the validity of alternative algorithms in school-aged children and adolescents. This study found that the frequency of PTSD-AA diagnosis (n=17; 29%) was slightly higher than the frequency of DSM-IV PTSD diagnosis (n=15; 25%). However, lowering the diagnostic thresholds did not result in overdiagnosis. Two children met the PTSD-AA but not the DSM-IV PTSD algorithm and both were highly symptomatic with a mean of eight symptoms, and reported significant functional impairment. Furthermore, PTSD-AA was more sensitive than DSM-IV PTSD in identifying children with functional impairment (.74 vs .63).

Excluding Criterion C3 from PTSD-AA did not affect prevalence rates in this study. However, the Criterion C threshold of one symptom appeared too inclusive as it had a 90% endorsement rate and failed to identify children with functional impairment. Excluding Criterion C3 improved the Criterion C validity as children with higher symptomatology and functional impairment were identified even when only one symptom was required. Based on the finding that two extra children with high symptomatology and impairment were identified, together with the findings of the utility of Criterion C thresholds and Item C3, it appears that PTSD-AA no C3 provides the most valid and developmentally sensitive criteria for identifying children who are highly symptomatic and functionally impaired. This finding requires replication in a larger sample, but together with Iselin et al.'s (2010) findings that PTSD-AA no C3 best identified children with psychosocial impairment following traumatic brain injury, it appears that this criteria may be the most appropriate for children when any organic alteration to consciousness is present during a traumatic event.

The results of this study also have implications for the recently published DSM-5 PTSD criteria. Changes in the new criteria include removal of DSM-IV Criterion A2 (‘the person's response involved intense fear, helplessness, or horror’). The conclusions of the current study that PTSD-AA no C3 is the most valid algorithm for this cohort supports this change in DSM-5. DSM-5 PTSD criteria also separate DSM-IV Criterion C into two clusters: Criterion C (avoidance) and Criterion D (negative alterations in cognitions and mood). While this study was unable to assess some of the new cognition/mood items, our findings do support a change to existing DSM-IV Criterion C, given that the previous threshold was not optimal for children. That the removal of DSM-IV Criterion C3 improved diagnostic validity in our cohort raises concerns about the use of this criterion (Criterion D1) in the DSM-5 with pediatric or adult samples with organic alterations in consciousness. Our results support the added stipulation in DSM-5 that the inability to recall key features of the traumatic event should reflect dissociative amnesia and should not be due to head injury, alcohol or drugs.

Limitations

Several limitations of this study should be acknowledged. The modest sample size and low participation rate reflect the challenges inherent in conducting research with this overburdened and critically ill cohort. In addition, the CPTSDI has been validated for children from seven years of age and the younger limit of our sample was six years. However, only two six-year-old children were included in this sample, and removal of their data did not substantively affect results. Finally, as this study was conducted before the DSM-5 PTSD criteria were proposed, we were unable to assess the prevalence and utility of new items and symptom clusters in this cohort. However, this study significantly contributes to our understanding of PTSD in school aged children and adolescents and has implications for assessment, diagnosis and treatment.

Conclusions

This study found that Scheeringa et al.'s (2006) PTSD-AA was the most valid and diagnostically sensitive algorithm for identifying distress and impairment in children aged 6–16. As in other cohorts with high endorsement of memory disturbance, DSM-IV Criterion C3 was not diagnostically useful and should be excluded from PTSD-AA in such cohorts.

Clinical Significance

This study has a number of implications for clinicians. Approximately one quarter of children in this cohort met criteria for PTSD. Clinicians should be aware that children are at high risk of significant psychological distress and impairment in everyday functioning up to six months following discharge from PICU. This study provides an understanding of the manifestation of PTSD in this cohort that may help to guide prevention and treatment.

Following substantial empirical evidence, the DSM-5 now features a PTSD Preschool subtype, recognising that developmental adjustments should be made to PTSD diagnostic criteria for young children. This study provides evidence that clinicians should also consider modifications from the DSM-IV criteria for school-aged children and adolescents. It appears that reliance on full DSM-IV criteria may fail to identify some children and adolescents suffering clinically relevant distress and impairment in everyday functioning. Utilizing some changes recommended for younger children (excluding DSM-IV Criterion A2 and reducing the Criterion C threshold) appears to facilitate more accurate diagnosis in older children (i.e., identify those suffering significant distress without being over-inclusive). Some of these changes appear in DSM-5 (e.g., removal of DSM-IV Criterion A2), but it remains to be seen whether the thresholds for the new Criterion C and Criterion D clusters will be adequate for school-aged children and adolescents. For specific populations of trauma-exposed youth with potential alterations to consciousness, clinicians also need to be mindful that endorsement of DSM-IV Criterion C3 (DSM-5 Criterion D1) may not represent psychological avoidance and thus should be excluded from diagnostic criteria in such cohorts.

Footnotes

Disclosures

Drs. Dow, Kenardy, Le Brocque, and Long have no conflicts of interest or financial ties to disclose.

References

American Psychiatric Association. Diagnostic, statistical manual of mental disorders, 5th. Washington, DC: American Psychiatric Association, 2013.

American Psychiatric Association. Diagnostic, statistical manual of mental disorders, 4th. Washington, DC: American Psychiatric Association, 2000.

Carrion

, Weems

, Ray

, Reiss

. Toward an empirical definition of pediatric PTSD: The phenomenology of PTSD symptoms in youth. J Am Acad Child Adolesc Psychiatry, 41:166–173. 2002.

Cohen

, Scheeringa

. Post-traumatic stress disorder diagnosis in children: Challenges, promises. Dialogues Clin Neurosci, 11:91. 2009.

De Young

, Kenardy

, Cobham

. Diagnosis of Posttraumatic Stress Disorder in preschool children. J Clin Child Adolesc Psychol, 40:375–384. 2011.

Dow

, Kenardy

, Long

, Le Brocque

. Children's post-traumatic stress and the role of memory following admission to intensive care: A review. Clin Psychol, 16:1–14. 2012.

Iselin

, Le Brocque

, Kenardy

, Anderson

, McKinlay

. Which method of posttraumatic stress disorder classification best predicts psychosocial function in children with traumatic brain injury? J Anxiety Disord, 24:774–779. 2010.

Levendosky

, Huth-Bocks

, Semel

, Shapiro

. Trauma symptoms in preschool-age children exposed to domestic violence. J Interpers Violence, 17:150–164. 2002.

Meiser-Stedman

, Smith

, Glucksman

, Yule

, Dalgleish

. The posttraumatic stress disorder diagnosis in preschool- and elementary school-age children exposed to motor vehicle accidents. Am J Psychiatry, 165:1326–1337. 2008.

10.

Ohmi

, Kojima

, Awai

, Kamata

, Sasaki

, Tanaka

, Mochizuki

, Hirooka

, Hata

. Post-traumatic stress disorder in pre-school aged children after a gas explosion. Eur J Pediatr, 161:643–648. 2002.

11.

Saigh

, Yasik

, Oberfield

, Green

, Halamandaris

, Rubenstein

, Nester

, Resko

, Hetz

, McHugh

. The Children's PTSD Inventory: Development and reliability. J Trauma Stress, 13:369–380. 2000.

12.

Scheeringa

, Wright

, Hunt

, Zeanah

. Factors affecting the diagnosis and prediction of PTSD symptomatology in children and adolescents. Am J Psychiatry, 163:644–651. 2006.

13.

Scheeringa

, Zeanah

. Reconsideration of harm's way: Onsets and comorbidity patterns of disorders in preschool children and their caregivers following Hurricane Katrina. J Clin Child Adolesc Psychol, 37:508–518. 2008.

14.

Scheeringa

, Zeanah

, Cohen

. PTSD in children and adolescents: Toward an empirically based algorithm. Depress Anxiety, 28:770–782. 2011.

15.

Scheeringa

, Zeanah

, Drell

, Larrieu

. Two approaches to the diagnosis of posttraumatic stress disorder in infancy, early childhood. J Am Acad Child Adolesc Psychiatry, 34. 1995.

16.

Scheeringa

, Zeanah

, Myers

, Putnam

. New findings on alternative criteria for PTSD in preschool children. J Am Acad Child Adolesc Psychiatry, 42. 2003.

17.

Slater

, Shann

, Pearson

. PIM2: A revised version of the Paediatric Index of Mortality. Intensive Care Med, 29:278–285. 2003.

18.

Stoddard

, Saxe

, Ronfeldt

, Drake

, Burns

, Edgren

, Sheridan

. Acute stress symptoms in young children with burns. J Am Acad Child Adolesc Psychiatry, 45. 2006.

19.

Yasik

, Saigh

, Oberfield

, Green

, Halamandaris

, McHugh

. The validity of the Children's PTSD Inventory. J Trauma Stress, 14:81–94. 2001.