Validation of the “Darryl” PTSD cartoon test with abused children

Research on trauma exposure and development of PTSD in children has increased. Studies have shown that most children are exposed to one or more potential traumatic events or childhood adversities during their lives (1, 2, 3). However, studies have also shown that a minor group of children are at risk of being exposed to several traumas and also certain types of traumas such as sexual, physical, and emotional abuse (4,5) and have high risks of developing PTSD later in life (6) and experiencing a range of other negative social, mental, and physical consequences (7). Similarly to studies of adult trauma survivors, a recent study found that a 78% co-morbidity between PTSD and other axis I disorders in a group of 80 traumatized and abused Danish preschoolers (8). The disorders included oppositional defiant disorder, separation anxiety disorder, and major depression as the most common. The endorsement of PTSD in this group was 46.3%.

Therefore, it is of most importance that clinicians can identify traumatized children as early as possible to allocate appropriate treatment. However, the younger the children, the more difficult it becomes to reliably assess reactions and symptoms following trauma exposure by child reports because children are still developing their cognitive, emotional, and behavioral skills. Many abused children have shown cognitive deficits. Cicchetti & Rogosch (9) found that abused children demonstrated inhibitions in talking about themselves and discussing negative emotions and thoughts. Also, many abused children may be less aware of negative feelings, more emotionally on guard, or have a great need for social approval (10). Post-traumatic reactions have included lower IQ and reading achievement (11), delayed language and poorer performance at school (12). Furthermore, children with PTSD have significant impairments in attention, abstract reasoning, and executive functioning compared to healthy controls (13).

This impacts how researchers and clinicians can ask questions in interviews or use questionnaires about some of the rather abstract symptoms of trauma, how children understand these questions, the children’s insight into own bodily, emotional, cognitive, and behavioral reactions, and the children’s ability to reliably answer such questions orally or by reading and checking scales or boxes in a written questionnaire. Pictorial cartoon tests accommodate some of these difficulties.

Two other cartoon tests for children's PTSD symptoms have been reported with validation evidence. The Cameron Complex Trauma Interview (CCTI) has the advantage of determining the potential traumatic event from the ten-item Adverse Childhood Experiences list introduced in the beginning and using a gender-neutral puppy. Eight psychological domains (attachment, biology, affect regulation, dissociation, behavioral control, cognition, self-concept, PTSD symptoms) are investigated by means of 21 symptoms on a 4-point visual Likert scale (with ‘no answer’ added as a 5th step) where the level of distress in Cameron’s face is used as indicator (14). This test has been used in one published study of 21 children collected through a snowball procedure. There is no reported investigation of the validity of the psychological domain items but clinicians’ ratings of using the test were positive. The clinicians were recommended to seek typical ‘who, what, where, when, how’ details for each of traumatic experiences and told “that at this point you do not need to inquire about emotional reactions”. The ACE part and the symptom parts correlated positively (r=.47), and both parts correlated well with the UCLA PTSD Reaction Index-5. The sample was small, and the selecting procedure biased. There have been no further studies using the test since the original study (15).

Another pictorial trauma test for young children is the Angie/Andy cartoon test (ACTS) that has been more thoroughly investigated. Four groups of children (intra-, extra-familial, combined, and non-trauma; n = 208) were compared on 22 PTSD items and 65 Complex PTSD items corresponding to the tentative DESNOS diagnosis in DSM-IV (16). There is no anchoring event; the index trauma must be determined before the start of the test. Angie/Andy was compared to a test, Child Rating Scales of Exposure to Interpersonal Abuse developed especially for this study as a research tool; the items being taken from the Conflict Tactics Scales (17) covering physical abuse, witnessing family and community violence. In addition, sexual violence was also covered. The article does not report the number of exposure items within the various groups of violence, nor is there any description of which items that has been assigned with a double weight. A parent rating scale was used to examine the concurrent validity of the test. The internal consistency was high, and the construct and concurrent validity was promising (18). The ACTS is no longer in print. For the time being due to lack of availability, lack of investigated construct validity, and procedural demands on extensive questioning about trauma exposure details, we do not consider the abovementioned tests viable alternatives.

A third cartoon test is “Darryl”, which was developed to examine cardinal symptoms of DSM-IV PTSD in school-aged children exposed to community violence (19). The Darryl includes both written questions of PTSD symptoms, which is read aloud by the interviewer to the child and accompanying pictures, which illustrate the preadolescent character, “Darryl”, who experiences each symptom after exposure to community violence. The child is then asked to rate how often they feel like Darryl. By using both auditory and visual stimuli during the assessment, the child’s attention and focus is maintained during a demanding task of talking about symptoms of trauma. The pictures assist in recognizing sensory elements from the child’s own abuse. The combination of pictures and a very short text read aloud followed by a simple question intends to provide a gentle, safe space for children to reveal their inner worlds. They can identify with another child so that they do not feel so alone with their feelings and symptoms. They can also point to the symptoms on the thermometers, which is a safe distance from themselves, as they do not have to verbalize. This approach is indirect, free of blaming, and non-threatening. The scoring choices are also adapted to a young child as they consist of three thermometers, one empty, one half full, and one filled, indicating whether the child ‘never’, ‘some of the time’ or ‘a lot of the time’ can relate to the symptom described in the picture (20).

Darryl was originally validated with a non-clinical sample of 110 American children aged 7-9 exposed to community violence (19). Internal consistency was found to be excellent for the overall scale of PTSD (α = .92) and acceptable to good for the subscales of reexperiencing, avoidance, and hyperarousal, respectively (α = .78, α = .83 and α = .80). Child reports of summed exposure scores were significantly associated with summed symptom scores on Darryl (r_s = 0.48, p <.001), as were adult reported exposure scores and the Darryl summed exposure scores (r_s = 0.29, p < .006). A second study with a clinical sample of 49 children and adolescents also found excellent overall internal consistency (α = .90) and adequate to acceptable internal consistencies for the three subscales (α = .72, α = .69 and α = .65). Also, the overall symptom score on Darryl was significantly correlated with the overall symptom score on the Child’s Post-traumatic Stress Disorder Reaction Index (CPTSD-RI, r = 0.64, p < 0.001) (21).

In Denmark, the original version of Darryl has first been translated and adapted for use in assessment of children who experienced a firework factory explosion in a residential area (22, 23). In addition, five different versions covering other types of traumas have been developed, including exposure to physical abuse, sexual abuse, war, disease, and death. Depending on the reason for referral to assessment, the corresponding version of the Darryl will be chosen. However, the different versions also call for studies examining the external validity of each version of the Darryl. Currently, the version used for exposure to a disaster (explosion) and exposure to violence have been validated with a sample of 430 children, aged 6-18, who were affected by the fireworks explosion (22) and a clinical sample of 65 children, aged 7-17 years, exposed to physical abuse (20). Both studies found good internal consistency for the overall scale of the Darryl (α = .89, (22); α = .88, (20). However, Elklit and colleagues (22) reported a Cronbach’s α of .59 for the subscale of hyperarousal, acceptable internal consistency for the subscale of reexperiencing (α = .71) and less than adequate internal consistency for the subscale of avoidance (α = .62). Internal consistencies for the version on physical abuse were adequate to acceptable (α = .78 for reexperiencing, α = .69 for avoidance, and α = .75 for hyperarousal) (20). Validity of the disaster version was examined using Darryl results from 215 children aged 12-18 years, where scores on Darryl and the Harvard Trauma Questionnaire (HTQ) and on Darryl and the Trauma Symptom Checklist (TSC) were both strongly and significantly correlated (r = .76, p < .01 and r = .74, p < .01, respectively) (22).

When the Darryl version for physical abuse was compared to the child and adolescent version of the Clinician Administered PTSD Scale (CAPS-CA) (24), scores were also strongly and significantly correlated (r = .73, p < 0.001) (20). The present study seeks to add to existing studies on the Darryl by examining the validity of two versions of the Darryl which are used in the Danish Children Centers (DCC), which assess children who are referred to the centers based on suspicion to exposure to physical and/or sexual child abuse (25).

The present study aims to examine reliability and validity of two versions of the Darryl cartoon test for use after exposure to physical abuse or sexual abuse. The first objective is to report PTSD symptomatology in children aged 6-17 years, who had been referred for psychological assessment due to suspected exposure to physical and/or sexual abuse. The second objective is to examine the reliability of the Darryl scale and subscales, and the third objective is to examine convergent validity of the Darryl when compared to another child-reported measure, the Beck Youth Inventories (BYI-II) (26-27) and when compared to the Strengths and Difficulties Questionnaire (SDQ) (28, 29) as reported by child caregivers. Thereby, this study provides further validation of the Darryl cartoon test and is the first study to examine reliability and validity of the version for assessment of children exposed to sexual abuse.

Of the 327 children who filled out the Darryl, a total of 238 children (126 boys and 112 girls) had been referred to a DCC on suspicion for exposure to physical abuse, while 58 children (15 boys and 43 girls) had been referred due to suspected exposure to sexual abuse. A minor group of 19 children (8 boys and 11 girls) were referred to the DCCs for suspected exposure to both physical and sexual abuse. For another group of 12 children, there is no report of the type of abuse. Table 1 shows the distribution of referral reason across gender.

This study set out to examine reliability and validity of the two versions of the Darryl cartoon test for use after exposure to physical abuse and sexual abuse. The study is the first to examine reliability and validity of the version for use after sexual abuse. Both versions performed well with a satisfactory reliability. The two abuse groups did not differ significantly on any of the Darryl scales but the group who had been exposed to both types of abuse (polyvictimization) scored remarkedly higher on all scales except the hypervigilant scale. This finding supports the criterion validity of Darryl scale, as polyvictimization generally in a dose-response way relates to higher levels of psychopathology and specifically to PTSD (31).

The convergent validity with the BYI-II was convincing with each of the five subscales correlating with the Darryl scales, strongest with the avoidance subscale and the total PTSD scale. The lesser associations with the disruptive and the self-concept subscales might reflect that the assessments are conducted shortly after the disclosure of the abuse and therefore disruptive behavior has not yet become as pronounced as it might after longer time if no preventive steps have been taken. At the same time, disruptive behavior is less common as depression and anxiety. The same more slowly developing deterioration can be said of the self-concept or self-worth which could be more devastated as time go by without any help.

The general lack of association between Darryl and the parent version of the SDQ is understandable when one considers that often the parent is not fully aware of the younger child’s problems due to cognitive and expressive limitations of the younger child. This caretaker underestimation can be conceived as a part of reciprocal protection in the family, which is very common due to the shame and taboos associated with child abuse. However, two subscales, the emotional and the impact scales had significant correlations with the avoidance and total PTSD scales. So, when the child reports avoidant symptoms and behavior, it seems related to the emotional problems registered by the parent and this impacts both the child and the family. Both with the BYI-II and the SDQ, the Darryl subscale of Avoidance seems to play a pivotal role linked to high scores on emotional problems and overall impact as well as more disruptive behavior and low self-worth. Although the analyses support both the reliability and the validity of the two Darryl versions for sexual and psychical abuse, there are several limitations to be aware of. First, recently abused children who are police interrogated and whose disclosure has caused alarm in their families and social network, are in an acute crisis. They must be treated gently and with compassion meaning that one should not exposed them to more challenges than absolutely necessary. This means that a ‘golden, perfect’ test situation with intensive and extensive interviews, repeated testing with the same instrument, or with several similar instruments is not possible at the DCCs. Therefore, we received the test data that the responsible clinical psychologist had chosen for a particular child without questioning the choices made.

Secondly, there is a dearth of trauma tests for young children. The CAPS-CA can in our experience be extremely stressful and potentially highly distressing due to its length and details for many seven years old children who are in a crisis but okay for most adolescents who are not in an acute crisis (20). Several questionnaires exist that are used with seven years old children, but very few, if any, build on a thorough examination of the child’s understanding of critical diagnostic trauma symptoms, the child’s understanding of time and grasp of time periods, or the child’s ability of express the degrees of symptom severity or frequency on a Likert scale with several or many steps.

The lack of a robust and simple validated Danish PTSD measure for very young children to assess the concurrent validity is a limitation in the current study. Future validation studies will have to rely on a clinical assessment of PTSD. This will take extensive training, more extensive parent or caregiver involvement, traumas that are not embedded in an acute or dramatic situation with possible dissolution of the family, and extended time for assessment compared to the present study.

The semi-projective methodology of the cartoon test with identification with a troubled child in a specific abuse or trauma context appeals to many young children as meaningful, and the translation from the plight of the figure to the child itself through pointing at a thermometer seems to work well even for very young children (six-seven years). The Darryl also seems to work well with older children and even adolescents in this specific assessment situation at the DCCs. The sample is relatively large and includes approximately equal numbers of each sex who have been victims of sexual or physical abuse who have been police interrogated and assessed by trained psychologist according to Danish law. The reliability for both versions of abuse were good and the convergent validity based on the BYI-II and the SDQ for the older and the younger age groups, respectively, was also good and meaningful.

For this young group of abused children, the development of a suitable measure with adequate psychometric properties is of paramount importance. Accurate assessment leads to the correct diagnosis and facilitates a focused clinical intervention that alleviates the symptom burden of the child. With the changes in diagnostic systems, a future task will be to make a validation of the various Darryl versions according to the ICD-11 (32) and the DSM-5 (33). Besides the change of the pictures and the text, the collection of data from this population of abused children will take some time because data collection is embedded in a clinical practice where also other authorities (e.g., the police, the courts, and the municipalities) are stakeholders. However, the data collection for the ICD-11 version of Darryl, that encompasses complex PTSD, is on its way. The ultimate quality of the Darryl is that it gives the young child a voice so that we can understand and react according to the symptoms and suffering the abused child reports.