Screening for Trauma Symptoms in Child Welfare-Involved Young Children: Findings from a Statewide Trauma-Informed Care Initiative

Abstract

Young children who experience abuse, neglect, domestic violence, and other interpersonal forms of trauma are at risk for developing complex psychological trauma. Timely referrals by child welfare services for trauma evaluation and intervention is critical, particularly during the developmentally-sensitive period of birth to three. However, few screening instruments exist that are feasible for implementation in child welfare services and none have reported psychometric data for children under three. The aim of this exploratory, retrospective study was to examine developmental differences in detection rates of two brief trauma screening scales, comparing outcomes for toddlers (age 1 and 2 years) and preschoolers (age 3 to 6 years), using the evaluation data from a statewide child welfare demonstration project. The sample included 151 children ages 1 to 6 participating in evidence-based trauma treatment with their caregivers. More than 80% of children, regardless of age group, met the cut-off on one of the screeners; children who met the cut-off on either screener were significantly more likely to have experienced domestic violence, physical abuse or poly-victimization. Implications for future research are discussed.

Children under the age of three are disproportionately represented in the child welfare system, collectively experiencing the highest rate of victimization compared with older children (U.S. Department of Health and Human Services 2017). Despite their young age, infants and toddlers involved with child protective services (CPS) have commonly been exposed to multiple potentially traumatic experiences, including the traumatic loss of being separated from their parent (Cooper et al. 2010; Osofsky et al. 2018). Neglect, abandonment, abuse, exposure to domestic violence, and other forms of complex psychological trauma can cause severe disturbance in the caregiver-child attachment relationship (Lieberman et al. 2011). When relational disturbance and dysfunction is chronic, and when it begins in the sensitive period of early childhood, development across multiple domains can be derailed causing a cascade of sequelae that affect physical health and emotional well-being across the lifespan (D’Andrea et al. 2012; DeBellis and Zisk 2014; Felitti et al. 1998).

Clearly, early identification of CPS-involved young children manifesting early symptoms of traumatic stress is critical for timely referral for further evaluation and trauma-specific intervention that can prevent negative sequelae in childhood, adolescence, and adulthood (Cohen and Scheeringa 2009; Conradi et al. 2011). As yet, the implementation of trauma screening in child welfare systems is in its early stages, with a number of states exploring various screening approaches to identify children’s exposure to potentially traumatic events and symptoms (Hanson et al. 2017; Lang et al. 2017). However, age-specific considerations for detecting trauma symptoms in the very youngest children is, at best, on the margins of these important systems improvement efforts due to both limited available screening options that are feasible for the child welfare setting as well as developmental challenges to screening.

Because very young children who are pre-verbal or early in their language, cognitive, and socioemotional development cannot report their symptoms, they must convey their internal emotional experiences through cues or behavior that can range from subtle to extremely intense and that can be difficult to interpret and to differentiate from normal reactions or from other early childhood disorders (Cohen and Scheeringa 2009; Lieberman et al. 2011). For example, infants and toddlers (i.e., children under the age of three years) experiencing traumatic stress may display separation fears and clinginess, exhibit aggressive behaviors or regressive behaviors, experience nightmares or sleep difficulties, startle easily, fear adults who remind them of the traumatic event, have a poor appetite and digestive problems, scream or cry excessively, or show lack of usual responsiveness (Zero to Six Collaborative Group 2010). In contrast, preschool-age children (i.e., 3 to 6 years of age) have more observable and discernible symptoms, in part because they have language to express their fears, worries, and report on physical symptoms. Traumatic stress symptoms among this age group include obvious anxiety and fearfulness; withdrawal and quieting; complaints of bodily aches and pains; a loss of focus and inattentiveness; re-enactment of the abusive/traumatic event in play; regressive behaviors such as bedwetting; and repeated retelling of traumatic event (Zero to Six Collaborative Group 2010).

Thus, screening tools for young children must reflect symptoms that are salient and age-specific for infants and toddlers as well as preschoolers. Moreover, if these tools are to be used in the child welfare system, symptom items must be observable to child welfare workers without specialized training as well as to parents or other caregivers who may not understand or recognize the expression of traumatic stress in young children. Several trauma exposure and trauma symptom assessment instruments appropriate for young children are currently available, with varying degrees of psychometric support (see Stover and Berkowitz 2005). As clinical tools, these measures are intended to be used in the context of diagnostic evaluation to collect information that will guide diagnosis, case formulation, and treatment planning or as clinical research tools. Accordingly, the requisite training and/or the length of time for administration precludes their feasibility as screening tools for child welfare workers to use to identify children needing further evaluation (e.g., Briere et al. 2001; Scheeringa and Zeanah 1994). While there are several tools designed explicitly for screening purposes, the few that are purportedly applicable for very young children span markedly wide age-ranges without psychometric data to support their reliability and validity with children under three years of age (Crandal and Conradi 2013).

Within this arena of limited screening options, two screening tools emerge as promising developmentally-appropriate approaches for very young children. The Young Child PTSD Screen or YCP-Screen is a very brief (6 item) screening tool derived from the Young Child PTSD Checklist (or YCP-Checklist), a clinical assessment measure for children ages 1 to 6 (Scheeringa 2010). The authors of the measure report that symptom items on the YCP-Screen were based on empirical research on trauma symptomatology in young children (Scheeringa and Haslett 2010), and the screener has demonstrated robust sensitivity with children in the 3- to 6-year-old age range; however, psychometric data on the YCP-Screen with children younger than 3 years is not yet available (Scheeringa 2010).

A second trauma symptom screening tool appropriate for young children (ages 1 ½ to 5 years) is the Child Behavior Checklist PTSD Scale (CBCL-PTSD), derived from the Child Behavior Checklist (Achenbach and Rescorla 2001). This screen comprises 15 items from the full measure and has been evaluated in several studies with children in the 2- to 6-year-old age range (see Rosner et al. 2012, for a description of the various ways the CBCL-PTSD has been operationalized). To date, several studies on the CBCL-PTSD have yielded promising though mixed evidence for the screening tool’s sensitivity (Dehon and Scheeringa 2006; Levendosky et al. 2002; Loeb et al. 2011). Levendosky et al. (2002) found a lack of concordance between a trauma assessment measure that the authors created, based on the DSM-IV, and the CBCL-PTSD scale. In contrast, both Dehon and Scheeringa (2006) and Loeb et al. (2011) found that CBCL-PTSD scores were significantly higher in children with a PTSD diagnosis based on the DSM-IV. However, the Loeb et al. study (Loeb et al. 2011) also found that children diagnosed with PTSD based on the Diagnostic Classification of Mental Health and Developmental Disorders of Infancy and Early Childhood (DC:0–3; Zero to Three 2005) did not have significantly different scores on the CBCL-PTSD compared with children without a DC:0–3 diagnosis, suggesting the CBCL-PTSD has limited sensitivity for identifying the full range of trauma symptoms in children under the age of three.

This article presents findings from an exploratory study examining trauma screening using data from the evaluation of a Federally-funded statewide demonstration project focused on increasing the capacity of child protective services to deliver trauma-informed services (the Massachusetts Child Trauma Project; see Bartlett et al. 2016; Barto et al. 2018; Fraser et al. 2014). The study investigates YCP-Screen and CBCL-PTSD outcomes, analyzed retrospectively, in a cohort of children receiving trauma treatment under the auspices of the Massachusetts Child Trauma Project. Specifically, we examined (1) the relationship between type and number of trauma exposures and screening outcomes, and (2) developmental differences in the screening tools’ detection in toddlers (ages 1 and 2 years) compared with preschoolers (ages 3 to 6 years). The study explores whether these screening tools are less sensitive in identifying toddlers at risk for traumatic stress compared with preschool-age children, given the subtle nature of trauma symptomatology in the younger age group. Additionally, the study sought contextual information for understanding screening outcomes by examining their association with type and pervasiveness of trauma (poly-victimization) experiences. Finally, as this is the first study to investigate the CBCL-PTSD screener with a sample of young children involved with the child welfare system and the first to examine the YCP-Screen with toddlers, the study aims to contribute to the sparse literature focused on screening with this population.

Method

Participants

The study sample comprised 151 young children (n = 106 children 3 years and older) and n = 45 children younger than 3 years, M_age = 4.02 years; SD = 1.40. The children’s ages ranged from 1 year (n = 6) to 6 years (n = 15). Children under 1 year of age were not included in the sample, as the YCP-Screen and CBCL-PTSD were not designed for use with infants under 12 months of age. The children had been referred by child welfare services for further evaluation and treatment services at community-based mental health agencies in the state’s child welfare system provider network, where clinicians were being trained in several evidence-based treatments (EBTs) under the auspices of the Massachusetts Child Trauma Project. Children and primary caregivers in the current study were participating in Child-Parent Psychotherapy (CPP), a treatment model developed specifically for trauma-exposed young children. CPP addresses children’s trauma through the vehicle of strengthening and repairing the parent-child relationship (Lieberman et al. 2015).

Clinicians at the participating agencies used a treatment selection protocol developed by the Massachusetts Child Trauma Project to assign children to the appropriate trauma-specific EBT. Children were assigned to CPP based on their age (0 to 6 years), caregiver involvement and availability to participate in treatment, and primary clinical presentation of predominant PTSD symptoms, behavior problems, or attachment difficulties. The sample is a subset of the full evaluation study population, which included children and youth up to age 17 years (see Bartlett et al. 2016; Fraser et al. 2014 for additional information on the evaluation design).

Table 1 presents demographic characteristics for the overall study sample and by age subgroup at intake. Most caregivers participating in the intake visit were birth parents, in both the younger and older age groups (67.7% and 56.4%, respectively). Younger children were somewhat more likely to have a foster parent participate compared with older children (24.4% vs. 18.9%, respectively), whereas older children were more likely to have another adult relative present (20.7% vs. 8.9%), although neither difference reached statistical significance. More children in the older age subgroup were Hispanic/Latino (34.0% vs. 19.1%); this difference also approached but did not reach significance. Most caregivers reported that English was the language spoken in the home (91.2%). Forty-eight percent of children in the sample were living at home with their birth parent, 22% with relatives or another family, 17.9% in regular foster care, 3.4% in treatment foster care, 2.8% in a shelter; the child’s primary residence was unknown for 4.8% of the sample. However, younger children were significantly more likely to be living in foster care compared with older children (χ² (1, N = 150) = 6.12, p = .013). The majority of children (N = 118; 81.9%) had not lived in their current caregiving setting for their entire life, regardless of age group.

Table 1.

Demographic characteristics

	Overall		< 3 Years		> 3 Years
	N = 151		n = 45		n = 106
Variable	N	%	n	%	n	%
Child Sex
Male	64	44.1	19	42.2	45	45.0
Female	81	55.9	26	57.8	55	55.0
Guardian
Birth Parent	69	47.3	21	46.7	48	47.5
Other Adult Relative	18	12.3	3	6.7	15	14.8
State	53	36.3	18	40.0	35	34.6
Other/Unknown/Missing	6	4.1	3	6.7	3	2.9
Respondent at Assessment Visita,b
Birth Parent	86	56.9	29	63.4	57	53.8
Other Adult Relative	26	17.2	4	8.9	22	20.7
Foster Parent	31	20.5	11	24.4	20	18.9
Race/Ethnicityb
White	117	86.0	36	87.8	81	85.3
Hispanic/Latino	41	29.5	8	19.0	33	34.0
Black/African American	13	9.6	4	9.0	9	9.5
Other/Unknown	6	4.4	1	2.4	5	5.3
Residence
Home/With Birth Parent	70	48.3	20	44.4	50	50.0
With Relatives/Other Family	32	22.1	6	13.3	26	26.0
Regular Foster Care	26	17.9	13	28.9	13	13.0
Treatment Foster Care	5	3.4	0	0.0	5	5.0
Shelter/Temporary Housing	4	2.8	1	2.2	3	3.0
Other/Unknown/Missing	6	4.1	5	1.1	3	3.0
Child Prescribed Medication	15	10.4	1	2.3	14	14.0
Types of Trauma Exposure
Domestic Violence	103	78.0	26	68.4	77	81.9†
Impaired Caregiver	136	71.3	28	70.0	69	71.9
Traumatic Loss	95	68.3	32	74.4	63	65.6
Neglect	75	55.6	24	60.0	51	53.7
Emotional Abuse	74	58.7	19	50.0	55	62.5
Physical Abuse or Assault	51	33.8	17	37.8	34	32.1
Sexual Abuse or Assault	16	10.6	0	0	16	15.1*
Extreme Interpersonal Violence	8	5.5	0	0	8	8.4†
Community Violence	7	5.4	0	0	7	7.7†

Table presents data collected at entry into treatment. N’s fluctuate by variable due to missing/incomplete data

^aDemographic and client history data was collected from other agency staff and collateral contacts for the few cases when a caregiver was not present for the baseline assessment visit

^bPercentages do not total 100 because of multiple respondents participating in the clinic visit and overlap across race and ethnicity

* p < .005; †p < .10

Referral Context

Increasing referrals of child welfare-involved children for trauma assessment and trauma-specific treatment was a goal of the Massachusetts Child Trauma Project. To facilitate referrals, the project provided training to child welfare workers to increase knowledge about children’s trauma symptoms; the mental health needs of children exposed to abuse, neglect, violence and systems-induced trauma (i.e., removal from the home, multiple placements); and effective trauma treatments. The project also provided support to communities in implementing teams of community stakeholders to increase linkages with and referrals to trauma-specific EBTs.

Although a formal trauma screening was not in place in the state’s child welfare system (Lang et al. 2017), the demonstration project disseminated an adaptation of the National Child Traumatic Stress Network (NCTSN) Child Welfare Trauma Referral Tool (Taylor et al. 2006) as part of its trainings with child welfare workers. The NCTSN tool was designed to help caseworkers organize information they have already gathered as part of the investigation to determine if a child needs to be referred for a trauma-specific treatment. The tool comprises questions about the child’s trauma exposure and reminders, traumatic stress reactions (re-experiencing, avoidance, numbing, and arousal), and internalizing and externalizing symptoms. Child welfare workers were encouraged to incorporate the tool into their practice, to enhance their capacity to make appropriate service referrals. A limited number of workers across the state voluntarily piloted the Massachusetts Child Trauma Project screen (Lang et al. 2017); data on child welfare workers’ screening practices and use of the tool was not collected as part of the demonstration project.

Recruitment

Children and caregivers were recruited as participants in the larger evaluation for the Massachusetts Child Trauma Project. Clinicians recruited children and caregivers for participation following assessment and enrollment in treatment services. Families were eligible to participate in the evaluation if they had involvement with the child welfare system within a year of referral and the children’s primary caregiver spoke English or Spanish. Children were recruited into the evaluation in three waves across a three-year period, as the demonstration project rolled out in different regions across the state.

Data Collection

Clinicians were trained to administer trauma exposure and symptom clinical assessment instruments at intake and at follow-up time-points or until treatment was complete or treatment was terminated. The data were entered into the project’s secure web-based data management system by staff at the community-based mental health agencies. For the purposes of this exploratory study, we analyzed only the baseline data. The evaluation was approved as exempt from review by the Human Subjects Institutional Review Board of the University of Massachusetts Medical School.

Measures

The Young Child PTSD Screen

As noted earlier, the YCP-Screen is a 6-item screener (see Table 2) derived from the Young Child PTSD Checklist (Scheeringa 2010), a clinical assessment instrument developed for assessing trauma symptoms in children age 1 to 3 years (see description below). The full clinical assessment was administered to caregivers, reporting on children’s symptoms, at intake into treatment services. The caregiver is asked to report on the degree to which each symptom has bothered the child using a Likert scale (0 = No, 1 = A little, 2 = A lot). The author of the scale reports that this Likert scale was created for administration purposes to give respondents a range of scores on the screener, in consideration of respondents possibly not endorsing mild or moderate symptoms if only a yes/no choice was given. Thus, any two endorsed symptoms (i.e., receiving a score of either 1 or 2) is considered a positive screen that should trigger a child’s referral for further assessment.

Table 2.

Young Child PTSD Screen

YCP-Screen items
1.	Does your child have intrusive memories of the trauma(s)? Does s/he bring it up on his/her own?
2.	Is your child having more nightmares since the trauma(s) occurred?
3.	Does s/he get upset when exposed to reminders of the event(s)? For example, a child who was in a car crash might be nervous while riding in a car now. Or, a child who was in a hurricane might be nervous when it is raining. Or, a child who saw domestic violence might be nervous when other people argue. Or, a girl who was sexually abused might be nervous when someone touches her.
4.	Has s/he had a hard time falling asleep or staying asleep since the trauma(s)?
5.	Has your child become more irritable, or had outbursts of anger, or developed extreme temper tantrums since the trauma(s)?
6.	Does your child startle more easily than before the trauma(s)? For example, if there’s a loud noise or someone sneaks up behind him/her, does s/he jump or seem startled?

On the full assessment measure, the respondent is asked to report on the frequency of child symptoms in the past two weeks, rated on a Likert scale of 0 = Not at all to 4 = Every day. For the current study, YCP-Screen scores were extrapolated from the assessment data, where any two endorsed screening items (i.e., items receiving a score of 1 or more) were considered a positive screen. The balance of re-experiencing and increased arousal symptoms on the YCP-Screen has shown excellent sensitivity (100%), positive predictive value (70.8), and negative predictive value (100%) when evaluated with preschool-age children (age 3 to 6 years). However, psychometric data is not yet available for the YCP-Screen with children under the age of 3 years (Scheeringa 2010).

The Child Behavior Checklist PTSD Scale

The CBCL-PTSD is a trauma symptom screening tool for preschoolers (Dehon and Scheeringa 2006; Loeb et al. 2011) that comprises 15 items from the CBCL (Achenbach and Rescorla 2001). Table 3 provides a list of the CBCL-PTSD items. The standard CBCL is a 113-item standardized caregiver report measure of child behavioral problems with high reliability as well as substantial concurrent and predictive validity with other parent report measures in clinical, non-clinical, and diverse samples (Achenbach and Rescorla 2001). Items are rated on a 3-point Likert scale. The study used the suggested cut-off of >9 (Dehon and Scheeringa 2006) to screen for greatest sensitivity in predicting children with a PTSD diagnosis. The CBCL is available in two versions, one for children ages 1.5 to 5 years and another for those ages 6 to 18 years. Studies reporting findings with the young child version of the CBCL-PTSD have reported moderate internal consistency for the CBCL-PTSD ranging from .79 to .83 (Dehon and Scheeringa 2006; Loeb et al. 2011). For the current study, children’s CBCL-PTSD scale scores were extrapolated from the full CBCL assessment measure.

Table 3.

Young Child CBCL-PTSD Scale

CBCL-PTSD items
1.	Defiant
2.	Cannot concentrate or cannot pay attention for long
3.	Clings to adult or too dependent
4.	Fears certain animals, situations, or places
5.	Nervous, high-strung, or tense
6.	Nightmares
7.	Too fearful or anxious
8.	Nausea and feels sick
9.	Stomachaches and cramps (without medical cause)
10.	Vomiting and throwing up (without medical cause)
11.	Stubborn, sullen, or irritable
12.	Sudden changes in mood or feelings
13.	Wakes up often at night
14.	Unhappy, sad, or depressed
15.	Withdrawn or does not get involved with others

Young Child PTSD Checklist Assessment

Children’s trauma symptoms and functional impairment were measured using the Young Child PTSD Checklist (YCP-Checklist), a 42-item clinical assessment instrument that includes 12-items assessing exposure to traumatic events, 24 PTSD symptom items, and 6 functional impairment items. As noted above, symptom frequency in the past two weeks is rated on a Likert scale (0 = Not at all to 4 = Every day) to generate a Total Symptom score The Symptom scale comprises three subscales: re-experiencing (7 items), avoidance and numbing (7 items), and increased arousal (5 items). For the Functional Impairment scale, caregivers indicate the extent to which the symptom “gets in the way” of child functioning. The author provides suggested cutoff scores of >26 for “probable diagnosis” and > 4 for “functional impairment” (Scheeringa 2010). The Symptom scale comprises three subscales: re-experiencing (7 items), avoidance and numbing (7 items), and increased arousal (5 items). The instrument has excellent face validity based on empirical research on trauma symptomatology in young children. The measure has demonstrated reliability and validity when the items were administered in the format of a diagnostic clinical interview (Scheeringa and Haslett 2010); however, there is not yet psychometric data available on the symptom items as administered in the assessment’s questionnaire format (Scheeringa 2010).

Core Clinical Characteristics: Trauma Exposure Measure

We assessed children’s exposure to trauma using a modified version of the clinician-administered Core Clinical Characteristics (CCC) measure developed by the National Child Traumatic Stress Network (NCTSN 2010). The measure asks respondents to report on whether the child experienced 18 specific trauma types with response choices were “yes,” “no,” and “unknown”; a modification in the present study is that the response option “suspected” was omitted. The form also asks clinicians to identify the primary focus of treatment. The present study included the trauma types most frequently reported by caregivers in the sample. Each trauma type was coded as not experienced (0) or experienced (1), with the total number of endorsements summed to represent cumulative trauma.

Data Analysis

We first examined type and number of trauma exposures across and within age group. Next, we calculated the proportion of children meeting the cut-off for the YCP-Screen and CBCL-PTSD, for the overall sample and within age groups. We also examined the YCP-Checklist (trauma assessment) scores for the children’s diagnostic status at intake. We calculated mean scores for the full sample and by age group for the Total Symptom, Symptom subscales, and Functional Impairment scale scores, as well as the percent of children in each age group who met the clinical cut-off on the measure’s Symptom and Functional Impairment scales. To elucidate the relationship between trauma exposure and positive identification on the screening tools, we examined the relationship between type of trauma exposure and screening cut-off, by age group, using Chi-square or Fisher’s Exact tests (when cell sizes were < 5). We report the phi coefficient (ϕ) to measure the effect size for Chi-square tests, where .1 is considered a small effect, .3 a medium effect, and .5 a large effect (Lewis-Beck et al. 2004). We compared the mean scores of younger vs. older children on the CBCL-PTSD and YCP-Checklist scores using t-tests (two-tailed, testing for possibility of a relationship in either direction); for the YCP-Screen, group difference was analyzed using the Chi-square test.

We assessed developmental difference in patterns of symptom endorsement on the screeners by examining the item-level distribution characteristics within each age group. In this analysis, we also included the full set of items on the YCP-Checklist, to assess items that may be more applicable for very young children (e.g., more clingy, new fears) that were not included in the screener. We first inspected the data visually, graphed as a histogram, for normality and then used the Shapiro-Wilk test to assess normality (Shapiro and Wilk 1965). We then conducted an item-level analysis examining mean, standard deviation, skewness, and kurtosis for each item on the measure. We identified patterns of symptom endorsement (i.e., items of which the majority of respondents did not endorse the symptom) as reflected in skewness and kurtosis statistics that represented a substantial departure from normality (i.e., skewness greater than 1 and kurtosis greater than 3). To assess developmental differences in patterns of endorsement further, we calculated the percentage of individuals with the lowest and highest possible score for each item to identify floor and ceiling effects. Data was missing for seven children for the YCP-Screen items and two children for the CBCL-PTSD items; missingness was attributable to the measure being omitted by the clinician for reasons unknown or to lack of entry into the database system. Participants with missing data were excluded from analyses. We conducted the statistical analysis using STATA/SE version 14.1 (Statacorp 2015).

Results

Table 1 presents data on trauma exposure for the overall sample and by age subgroup. The younger and older age groups did not differ in cumulative trauma exposure (M = 3.31; SD = 0.29; Range 0–7 vs. M = 3.73; SD = 0.18; Range 0–7, respectively). Across and within age groups, more than two-thirds of children had been exposed to domestic violence, an impaired caregiver (due to substance addiction or other parental mental health problems), and/or traumatic loss. None of the younger children were known to have experienced sexual abuse or assault and the older children were more likely to have experienced other forms of violence, although this difference did not reach statistical significance.

Across the sample, children who met the cut-off on the YCP-Screen items and the CBCL-PTSD scale were significantly more likely to have experienced multiple traumas compared with children who did not meet the cut-off, M = 3.92 (SD = .15) vs. M = 1.61 (SD = .30), t(148) = 5.69, p < .0001, and M = 3.79 (SD = .18) vs. M = 2.99 (SD = 2.21), t(151) = 2.88, p < .005 respectively. Regardless of age group, children who met the cut-off for trauma screening on the CBCL-PTSD or YCP-Screen items were significantly more likely to have experienced domestic violence, (χ² (1, N = 151) = 9.65, ϕ = .22, p = .002, and Fisher’s Exact, N = 132, ϕ = .33, p = .001, respectively, and physical abuse or assault, χ² (1, N = 151) = 8.65, ϕ = .18, p = .003, and Fisher’s Exact, N = 132, ϕ = .25, p = .002, respectively.

There was no significant difference between younger and older children in identification on the YCP-Screen, with 84.4% and 86.8% of children meeting the cut-off, respectively. Cut-off rates for younger compared with older children on the CBCL-PTSD scale were again similar but markedly lower than the YCP-Checklist-Screen results (46.7% vs. 54.7%, respectively). Although not reaching significance, a difference emerged across type of respondent (i.e., caregiver at the clinical visit), with 100% (n = 20) of other adult relatives and 95% (n = 23 out of 24) of foster parents reporting symptoms that met the cut-off on the YCP-Screen items compared with 15% of birth parents (n = 65).

Results on the full assessment measure (the YCP-Checklist) showed a pattern of lower mean scores for younger children compared with children in the older age group (see Table 4). Regardless of age group, and for the sample overall, approximately 30% or fewer children met the cut-off on the Symptom scale: 33.9% of older children, 24.4% of younger children, and 31.1% of the overall sample. The difference in symptom assessment by age group approached but did not reach significance. In contrast, children were more likely to have clinically elevated scores on the Functional Impairment scale, with 70.2% of the overall sample, 75.5% of older children, and 57.8% of younger children meeting the clinical cut-off. However, preschool-age children were significantly more likely to meet the clinical cut-off on the Functional Impairment scale compared with toddlers (χ² (1, N = 151) = 4.73, p < .05).

Table 4.

Mean, standard deviation, and range on measures

	Total Sample	< 3 Years	> 3 Years
	M (SD)	M (SD)	M (SD)
	Range	Range	Range
CBCL-PTSD	8.76 (6.18)	8.24 (6.10)	8.99 (6.24)
	0–28	0–23	0–28
YCP-Checklist Symptom Scale	21.73 (16.56)	19.18 (15.85)	22.82 (16.81)
	0–65	0–63	0–65
YCP-Checklist Re-experiencing Subscale	5.89 (5.33)	5.33 (5.13)	6.13 (5.42)
	0–20	0–20	0–65
YCP-Checklist Avoidance/Numbing Subscale*	4.08 (4.45)	2.52 (3.94)	4.78 (4.51)
	0–19	0–18	0–19
YCP-Checklist Increased Arousal Subscale	7.26 (5.11)	6.26 (4.90)	7.69 (5.16)
	0–20	0–16	0–20
YCP-Checklist Functional Impairment Scale*	7.82 (6.71)	5.45 (5.86)	8.69 (6.89)
	0–24	0–23	0–24

Screening cut-off for CBCL-PTSD >9. Probably Diagnosis cut-off: YCP-Checklist Symptom Scale >26, YCP-Checklist Re-experiencing Subscale >8, Avoidance/Numbing >4, Increased Arousal >10, Functional Impairment Scale >4

*p < .05

Results of the Shapiro-Wilk test indicated that the CBCL-PTSD was not normally distributed: S-W = 0.970, df = 151, p = .003. Four items were heavily positively skewed (i.e., a higher frequency of not endorsing statements describing symptoms) for both younger and older children: three somatic items (nausea, stomachaches, vomiting) and ‘withdrawn.’ For each of these items, more than 70% of respondents did not endorse the symptom. One additional item, ‘unhappy/sad/depressed,’ was positively skewed only for the younger children.

Table 5 presents item-level findings for the YCP-Screen item and YCP-Checklist. Among the 6 items on the screener, two were heavily skewed but for the younger children only (‘intrusive memories’ and ‘startle more easily’). On the full assessment measure, results of the Shapiro-Wilk showed significant skew: S-W = 0.922, df = 151, p = .001. Two-thirds of items (67%; n = 20) were markedly positively skewed for the younger age group, with endorsement of “0” on items ranging from 50.0% to 90.9% of respondents. Symptom items that were not skewed for toddlers, other than those included on the YCP-Screen, were ‘more physically aggressive’ and ‘more clingy.’ For preschool-age children, positively skewed items for which more than 50% of respondents endorsed ‘0’ was less prevalent (40% of symptom items), with no skewness and greater response variability for the functional impairment items.

Table 5.

Item characteristics YCP-Checklist by age subgroup

		Age < 3				Age > 3 Years
		Mean	SD	Skewness	Kurtosis	Mean	SD	Skewness	Kurtosis
Symptom Scale
1	Intrusive memories	0.57	1.04	1.93	6.15	1.13	1.21	0.98	3.08
2	Re-enact trauma in play	0.34	0.64	2.19	8.24	1.08	1.24	0.85	2.57
3	More nightmares	1.20	1.44	0.96	2.55	1.05	1.24	0.98	2.85
4	Act like event is happening again	0.52	1.04	2.16	6.92	0.43	0.94	2.46	8.48
5	Freeze/unresponsive	0.77	1.09	1.53	4.67	0.60	1.12	1.98	5.99
6	Upset when exposed to reminders	1.26	1.27	0.79	2.73	1.22	1.18	0.86	2.99
7	Physical distress	0.50	0.90	2.11	7.43	0.63	1.02	1.63	4.75
8	Avoid conversations	0.53	1.12	2.22	6.95	0.95	1.24	1.19	3.35
9	Avoid reminders	0.46	1.08	2.58	8.63	0.65	1.19	1.81	5.11
10	Difficulty remember incident	0.37	1.00	2.81	9.78	0.82	1.24	1.48	4.13
11	Lost interest in things	0.11	0.39	3.55	15.44	0.29	0.85	3.01	11.31
12	Restricted range of emotions	0.31	0.83	3.07	12.34	0.57	1.02	1.80	5.08
13	Lost hope for the future	0.16	0.65	5.05	29.43	0.43	1.02	2.36	7.27
14	Become more distant/detached	0.48	0.87	1.85	5.44	0.73	1.31	1.65	4.24
15	Hard time falling asleep/staying asleep	1.81	1.65	0.14	1.42	1.72	1.66	0.26	1.40
16	More irritable, angry outbursts, extreme temper tantrums	1.93	1.39	0.12	1.81	2.06	1.44	0.10	1.67
17	More trouble concentrating	0.70	1.27	1.68	4.56	1.37	1.47	0.65	2.00
18	More “on the alert” for bad things to happen	0.89	1.45	1.45	3.49	1.22	1.49	0.93	2.36
19	Startle more easily	0.86	1.27	1.43	3.95	1.26	1.34	0.74	2.29
20	More physically aggressive	1.56	1.50	0.61	1.95	2.05	1.47	−0.29	1.65
21	More clingy	1.95	1.52	0.12	1.56	2.11	1.62	−0.11	1.41
22	Night terrors start or get worse	0.84	1.24	1.49	4.27	0.69	1.16	1.84	5.43
23	Lost previously acquired skills	0.65	1.38	1.81	4.52	0.53	1.02	2.05	6.32
24	New Fears	1.0	1.40	1.14	2.92	1.22	1.43	0.74	2.10
Functional Impairment Scale
25	Symptoms affect relationship with parent/caregiver	1.34	1.44	0.79	2.29	1.62	1.45	0.42	1.80
26	Symptoms affect relationships with siblings	1.09	1.51	0.96	2.35	1.48	1.54	0.53	1.76
27	Symptoms affect teacher or class	0.45	1.62	2.44	8.10	1.17	1.41	0.93	2.51
28	Symptoms affect friendships	0.68	1.27	1.78	4.82	1.21	1.41	0.94	2.52
29	Hard to take him/her out in public	0.98	1.28	1.24	3.43	1.55	1.58	0.47	1.66
30	Behaviors cause child to feel upset	1.41	1.42	0.63	2.11	1.78	1.59	0.20	1.48

YCP-Screen items are italicized. Italicized numbers indicate skewness >1.0 and kurtosis >3.0. Re-Experiencing Subscale: Items 1–7; Avoidance/Numbing Subscale: Items 8–14; Increased Arousal Subscale: Items 15–19

Positively skewed responses were also notable for the younger children on all three subscales on the assessment measure. Among older children, fewer items were skewed on the Re-experiencing subscale and none of the Increased Arousal items were skewed; however, all items were skewed on the Avoidance/Numbing subscale.

Discussion

This exploratory study aimed to contribute to the research literature on appropriate trauma screening approaches for very young children, using data from a real-world child welfare system improvement effort. The study applied a developmental lens to further our understanding of screening for trauma symptoms in a sample of very young children receiving child welfare services, all of whom had been identified as needing treatment based on clinical presentation at intake into mental health services. Retrospective analysis of YCP-Screen items found that more than 80% of children, regardless of age group, met the cut-off for needing further trauma assessment. However, a high detection rate was found only for the YCP-Screen compared with the CBCL-PTSD. The comparatively lower number of children who met the cut-off on the CBCL-PTSD suggests that the items miss trauma symptoms that are most salient for young children. Alternatively, perhaps they were particularly difficult for caregivers in a child welfare population in which foster parents, other adult relatives, and birth parents involved with CPS were respondents. Regardless of screener, parents in the study were less likely to endorse children’s trauma symptoms compared with foster parents and other adult caregivers. It may be that parents may not perceive, understand, or notice small changes in their children’s emotional state or behavior, particularly if they are struggling with their own trauma symptoms or mental health issues, substance use disorders, or living in families and communities where trauma and adversity is the norm. Parents also may minimize, either consciously or unconsciously, harms their children have experienced out of guilt and shame.

The study findings suggest the very brief YCP-Screen can detect signs of possible traumatic stress in very young children and identify children in need of further clinical evaluation; however, the screener did not capture all children who had been identified through clinical intake as needing trauma treatment. This may be attributable to the absence of any items on the screener inquiring about attachment difficulties (e.g., separation fears, clinginess, indiscriminate attachment behavior, lack of responsiveness).

Limitations

This study has a number of important limitations. First, the exploratory and largely descriptive design of the study prevents our drawing definitive conclusions. Additionally, we could not control for variables that may have contributed to the screening or clinical findings, including those differences that were significant (or approaching significance) between the groups. To examine the items on the YCP-Screen, we retroactively extrapolated data from the YCP-Checklist assessment measure. Thus, the generalizability of the findings is limited by the possibility that caregivers may respond differently to screening questions in the child welfare context as opposed to the clinical context of a mental health treatment setting. We also could not perform analyses of the psychometric properties of the screeners to assess sensitivity, positive predictive value, and other psychometric indices in relation to an external criterion of an independent assessment of young children’s traumatic stress. The evaluation did not include a structured clinical diagnostic interview appropriate for very young children using the DC: 0–5™ (Zero to Three 2016), which would have allowed rigorous psychometric analysis. In this regard, our findings should be considered provisional as future research is needed to ascertain whether the YPC-Screen over-identifies or the CBCL-PTSD under-identifies children with clinically significant traumatic stress problems and how their performance varies by age.

We must caution, as well, that the lack of a balanced representation of racial and ethnic diversity in the sample, with comparatively few African American children compared to White and Hispanic/Latino children, is a significant limitation affecting the generalizability of the study findings. The underrepresentation of African American children in the sample is notable, given their disproportionate representation in the child welfare system and less equitable service receipt (Child Welfare Information Gateway 2016). Implicit bias may have influenced referral decision-making and it may also be that Black families were less likely to engage in CPP treatment due to structural barriers, issues of trust regarding the child welfare system, and the stigma of mental health services.

Implications

The current study provides preliminary data suggesting promising findings for the YCP-Screen with very young children in the child welfare system that points to areas for future research. Specifically, future studies may investigate whether adding items from the YCP-Checklist that are highly developmentally salient for toddlers (‘more clingy’ and ‘new fears’), and which caregivers were more likely to endorse in this study, improves the YCP-Screen’s performance. Future, more methodologically rigorous research should also assess item performance, using item response theory models or factor analytic models to formally test if the measures are operating differentially for different age groups. Future research also should ensure a representative sample and explore the role of race and ethnicity on screening outcomes. The study results also provide further support for previous research that calls into question the use of the CBCL-PTSD as a trauma screen with very young children. Additionally, the study, not surprisingly, found that children exposed to domestic violence, physical abuse, or poly-victimization were highly likely to be identified as needing further evaluation for signs of traumatic stress. Clearly, these trauma exposures may warrant triaging for symptom screening or outright referral, based on exposure alone.

The findings that younger and older children differed primarily on the Avoidance/Numbing and Functional Impairment scales of the YCP-Checklist are not surprising, given differences in emotional development and different types of functioning in toddlerhood versus the preschool years. These findings do, however, call attention to the limited number of items on this assessment measure that are salient to traumatic stress symptoms in 1-year-olds and 2-year-olds. Combined with the finding that less than 25% of the toddlers in the sample met the clinical cut-off on the YCP-Checklist, the study points to the need for future psychometric research to improve the measure’s applicability for this younger group.

Young children’s symptomatology is particularly complex as it is expressed across multiple domains of child functioning and in nuanced ways (Cohen and Scheeringa 2009; Scheeringa et al. 2004). Moreover, the challenge of caregiver report in assessment is amplified in the child welfare system, where foster parents and adult relatives may have a limited history with the young child and the birth parent’s reflective capacity may be impaired by his or her own trauma or other mental health issues. For these reasons, a comprehensive evaluation process that assesses children’s emotional, social, and developmental functioning and includes assessment of child-parent relationship functioning and direct observation of play and child-parent interaction is essential for accurate diagnosis and effective treatment (Lieberman and Van Horn 2007).

As awareness of the prevalence and impact of trauma on children involved with CPS grows, and trauma-informed child welfare practice that includes identification and referral to trauma-specific treatment continues to spread (Chadwick Trauma-Informed Systems Project 2013), the need for psychometrically valid brief trauma screening tools that are relevant for child welfare services is increasingly urgent. Though preliminary and exploratory, the results of this study provide rich information about the applicability of brief screening tools to the intended target population. The study also suggests direction for future research on trauma screening for very young children, with the goal of enhancing the capacity of child welfare systems to ensure vulnerable children receive appropriate and timely intervention that can prevent future misdiagnoses, worsening of symptoms, and a trajectory of poor outcomes across the life course.

Funding

The study was supported by a grant from the Administration for Children and Families, Children’s Bureau (Grant No. 90C01057) to the Massachusetts Department of Children and Families.

Compliance with Ethical Standards

Conflict of Interest

The authors declare that they have no conflicts to report.

Ethical Standards and Informed Consent

All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation [institutional and national] and with the Helsinki Declaration of 1975, as revised in 2000. Informed consent was obtained from all patients for being included in the study.