PTSD in Children Younger Than the Age of 13: Toward Developmentally Sensitive Assessment and Management

Abstract

Assessment of posttraumatic stress disorder (PTSD) is more difficult compared to other disorders for multiple reasons that are listed and explained. Multiple causes of low sensitivity for making the PTSD diagnosis in children are discussed. Diagnostic validity and comorbidity issues in particular are highlighted. For the diagnostic criteria, wording changes to five of the items have been proposed, but the most substantial proposed change is lowering the criterion C requirement from three to one symptom. Early studies suggest the course is more chronic. Parenting effects are reviewed and caution is urged before drawing premature conclusions about the directionality of effects. Advice for interviewing respondents is organized into seven practical suggestions. Treatment implications from the above are discussed.

Since posttraumatic stress disorder (PTSD) was codified (American Psychiatric Association, 1980) there has been some controversy in general about the diagnosis, and some skepticism about the existence of the diagnosis in children (Benedek, 1985), but this is no longer debatable. While previous iterations of Diagnostic and Statistical Manual (DSM) criteria for PTSD were published throughout three editions with few data available from children less than 18 years old and almost no data on children less than 12 years old, much research has been conducted in the past 15 years to establish the diagnostic validity of PTSD in preschool and school-age children. But detection of PTSD in children in clinic settings is time-consuming and difficult. The challenges for assessment of infant, preschool, and school-age children have only recently been described and have far to go in terms of adequate application in clinic settings.

Assessment of PTSD is Relatively More Difficult Compared to Other Disorders

The current diagnostic criteria for PTSD require that children have experienced, witnessed or learned of a traumatic event, defined as one that is terrifying, shocking and potentially threatening to life, safety or physical integrity of self or other. Children must also meet at least one reexperiencing criteria, three avoidant/numbing criteria, and two increased arousal criteria. Children must meet minimal duration criteria of at least one month, and they must show functional impairment in an important area (school, peers, family, etc). This is a difficult threshold to meet relative to most other disorders for at least seven reasons.

1. PTSD is one of the rare disorders in the DSM-IV that requires an etiological component. Before asking about any symptoms of PTSD, one must go through a menu of all possible traumatic experiences because many children have experienced more than one traumatic event. In a community representative sample of 1,420 children, 37.0% had been exposed to more than one traumatic event (Copeland, Keeler, Angold, & Costello, 2007). The need to inventory exposure to trauma is time-consuming and relatively invasive and may run counter to efforts to build rapport.

2. There are 17 possible symptoms in the PTSD DSM-IV criteria. This is more than double the number found in nearly all other disorders, with the exception of attention-deficit/hyperactivity disorder (ADHD) and conduct disorder. At a practical level, this more than doubles the amount of time to conduct an assessment per disorder. For this article, data were extracted post hoc from an existing study to document how long it takes to interview for PTSD compared to other common disorders. In an assessment study funded by the National Institute of Mental Health, 3–6 year-old trauma-exposed children were assessed with the Preschool Age Psychiatric Assessment (Egger et al., 2006), an interview of caregivers about children. Twelve interviews were sampled in a quasi-random selection process. The durations required for each disorder were 52.4 minutes for PTSD, 25.6 minutes for major depressive disorder, 13.3 minutes for ADHD, 10.3 minutes for oppositional defiant disorder (ODD), 8.0 minutes for phobias, 6.5 minutes for separation anxiety disorder (SAD), and 4.3 minutes for generalized anxiety disorder.

3. Every one of the 17 symptoms has to be tied back to an event/date of onset. It is insufficient to establish that only one of the 17 PTSD symptoms developed immediately after a traumatic event. This process is, again, time consuming, and also poses unique technical challenges. Many of these inquiries require multi-step connections for both the interviewer and respondent: identify the symptom, connect it back in time to a past event, and recognize that the current manifestations are similar to the past experiences.

4. Memory is a unique and key aspect of PTSD not present in any other disorder. Memory of a traumatic event is required for the disorder, and memory to connect present symptoms to past experiences is required to discuss symptoms in an assessment. Yet, these memories and symptoms are, by definition, unpleasant. The interviewer attempts to solicit descriptions of events the respondent would like to forget. This issue is closely tied to two of the seven symptoms in criterion C (avoidance and numbing symptoms) that are about avoidance of reminders of past events. There is especially the challenge of asking respondents to report avoidance symptoms because the nature of the symptom is to actively avoid talking or thinking about it (Cohen & Scheeringa, 2009). Respondents can be prone to reply, “I don’t want to talk about it” or “I don’t think about it.” The former response puts up a roadblock that sensitive clinicians fear to cross and the latter response sounds like a denial when it may actually be an endorsement. In each case, the response contains insufficient information. A clinician must probe further to determine if any of these children may be actively avoidant and may have significant PTSD symptoms. Arguably, patients with other types of disorders also find it unpleasant to discuss their symptoms, but the inherent embeddedness of unpleasant memory to PTSD is unlike any other disorder.

A complication to this challenge is a basic misunderstanding of PTSD symptomatology. Oftentimes, avoidance has been dismissed (incorrectly) as adaptive learning from experience. The reasoning goes that if children incurred harm from past experiences (say, hit by a car while crossing a street), and subsequently act fearful when placed in a similar situation that reminds them of the past event for them (e.g., while crossing streets), then it is only logical and appropriate (i.e., it is not a symptom). What this assertion fails to acknowledge is that caution is a normative reaction for children when crossing busy streets, but fear is not. The spirit of the PTSD symptom is that new distress is coupled with the avoidant behavior, not simply acting with normative apprehension.

5. Many of the PTSD symptoms are highly internalized, which makes them difficult for parents to observe and difficult for children to express with developing language and narrative capacities. Internalized symptoms include avoidance of internalized reminders of past traumas, inability to recall events (due to psychogenic blocking of painful memories), sense of a foreshortened future, feeling of detachment or estrangement from people, certain types of psychological distress to reminders of traumas, physiological reactivity to reminders of traumas, dissociative experiences, intrusive recollections, and perhaps nightmares. In addition to the obvious validity problem of limited detection accuracy, this presents a practical challenge in that it often requires more probing to assess these symptoms comprehensively, which takes more time.

6. Because most persons have never had PTSD, caregivers may have no common frame of reference from which to report on their children’s feelings. Contrast this to other types of disorders, such as depression or ADHD or ODD. Nearly everyone intuitively understands and recognizes sadness, hyperactivity, and defiance; these are easily observable and require little in the way of verbalizations and self-recognitions from children for caregivers to be aware of them. A lack of familiarity with PTSD symptoms adds to the time and complexity of the interview process because a proper interview must educate the respondent about each symptom while concurrently asking about each symptom.

7. Many PTSD symptoms manifest differently at different developmental stages. For example, the symptom of intrusive recollections is manifest by school-age children almost entirely by verbalizations, but is manifest by many preschool-age children as reenactments in their play that do not depend so heavily on verbalizations. Avoidance behaviors, restricted range of affect, diminished interest in activities, detachment from loved ones, difficulty concentrating, and irritability/outbursts of anger may also have systematic differences in presentations by age. This requires awareness on the part of the interviewer of developmental differences and flexibility in interviewing. A lone stock question is not sufficient for all age periods or even for one child. Multiple ways of asking about a symptom may be needed, which, again, requires time, and effort.

Diagnostic Validity, Sensitivity, Specificity, and Comorbidity: Evidence for PTSD as the Core Trauma Response

The tenor of past controversies around PTSD seem to have been extreme because PTSD is one of the rare disorders with an etiologic event, and because many claims for diagnoses have been in the context of claims for disability, financial gain, and legal manipulations. But ignoring those diversions, the case for the validity of PTSD as a distinct diagnostic construct is as strong as for any other psychiatric disorder. In fact, PTSD is one of the most well-studied and validated disorders in longitudinal, neurobiological, and treatment response studies. Some clinicians, scholars, and other observers may be dissatisfied with the complexity and messiness of posttraumatic responses, but the data do not support a wholesale deconstruction of PTSD based on false negatives or false positives. Not all of the diagnostic validity evidence can be reviewed in this space, but several major issues of practical clinical assessment are reviewed.

For a larger context in which to view assessment of children, it is worth noting that the most frequent criticism about PTSD in adults is a lack of specificity; concerns that persons who do not really have the disorder may be diagnosed (too many false positives). For example, in one study of patients enrolling in treatment studies for depression, the group with true trauma events and the group with “minor traumas,” both had nearly 80% rates of PTSD from structured interviews (Bodkin, Pope, Detke, & Hudson, 2007). That is, the “minor trauma” subjects endorsed enough criteria B, C, and D symptoms from their events to qualify for the diagnosis. However, the authors paid only glancing attention to the issue that this was a highly selective help-seeking depressed sample that may have had greater vulnerability to react to minor trauma and develop symptoms. Furthermore, no attempt was made to comparatively grade the severity of PTSD within each of those groups to explore the possibility that the “minor trauma” group may have had relatively less severe PTSD than the trauma group.

Another fact that drives some concerns about specificity is that overlap exists between diagnostic criteria for PTSD and other internalizing disorders. Four PTSD diagnostic criteria (decreased interest in activities, sleep disturbance, restricted range of affect, and decreased concentration) overlap with MDD. Three symptoms of PTSD (decreased concentration, irritability, and sleep disturbance) overlap with GAD.

Speculating from concerns about overlap and studies such as Bodkin et al, Spitzer and colleagues proposed modified diagnostic criteria for PTSD in an effort to restrict who can receive the diagnosis (Spitzer, First, & Wakefield, 2007). The main suggestion was to eliminate five symptoms that overlapped with other disorders. In addition, the requirements for three out of seven symptoms from criterion C plus two out of five symptoms from criterion D were replaced with a single criterion (criteria C and D collapsed) of seven possible symptoms, of which four symptoms were required. Unfortunately, these changes were proposed in the absence of empirical data. Elhai and colleagues reviewed the data of 5,692 participants in the National Comorbidity Survey Replication and found that these recommendations made an insignificant impact (Elhai, Grubaugh, Kashdan, & Frueh, 2008). The recommendations lowered the rate of PTSD only from 6.81% to 6.42%. These authors concluded that “little difference was found between the criteria sets in diagnostic comorbidity and disability, structural validity, and internal consistency” (p. 597).

In contrast, the prominent concern in the child literature has not been about specificity because historically the issue “in the clinical trenches” is that children have been under-recognized as having internalizing symptoms, not over-diagnosed. In other words, the concern has been lack of sensitivity rather than lack of specificity. When PTSD was first being recognized in adults, disbelief was expressed at the notion that children could also manifest PTSD (Benedek, 1985). This phenomenon could have been due to misunderstandings on the part of adult clinicians about the true developmental capacities of children. But even among trained child clinicians, recognition of PTSD in clinics can be lacking or slow. One can speculate that this is partly because one cannot look at a child and tell that he/she has PTSD. Returning again to the analogy of the more observable disorders of ADHD, ODD, and MDD which one can fairly obviously see in a mental status exam (or a classroom, or a living room), PTSD is relatively less obvious. The key and pathognomonic symptoms are triggered by trauma reminders and can be uncommon. And among other professionals and parents, even when “something is wrong” is recognized, the obviously new onset tends to make one believe that gradual offset will naturally follow.

PTSD is also challenged by claims that children who have been chronically exposed to interpersonal trauma in early childhood either cannot qualify for PTSD or need a better diagnosis than PTSD (van der Kolk, 2005). Or, when they are diagnosed with PTSD plus the inevitable comorbid disorder(s), this purportedly misleads clinicians to treat comorbid conditions rather than the trauma syndrome and “may run the risk of applying treatment approaches that are not helpful” (p. 406). A new syndrome was proposed, developmental trauma disorder but without preliminary validity data (van der Kolk, 2005). At this point, it has not yet achieved face validity. Interestingly, no one has documented that the signs and symptoms in chronically and repeatedly traumatized youngsters are not adequately represented by PTSD.

Underlying these concerns about sensitivity and specificity is no doubt the high degree of comorbidity of other disorders with PTSD. In both adult and child populations 80–90% of the time PTSD occurs with at least one other disorder. Implicit in the arguments of Spitzer et al., is that comorbidity is clouding the picture, specifically, that non-PTSD symptoms are being misidentified as part of PTSD because they overlap. In adults, the common comorbid conditions include depression, anxiety, and substance abuse (Kessler, Sonnega, Bromet, Hughes, & Nelson, 1995). In preschool children, the most common comorbid disorders are ODD and SAD. In one study with 1–6 year-old children who experienced heterogeneous types of trauma events, the comorbid frequencies with PTSD were 75% for ODD and 63% for SAD (Scheeringa, Zeanah, Myers, & Putnam, 2003). In a second study when all victims suffered a hurricane disaster, the comorbid rates were 61% for ODD and 21% for SAD (Scheeringa & Zeanah, 2008). The issue in older children and adolescents is less well-documented.

Nevertheless, there are at least three reasons why comorbidity is neither a dense conundrum nor an issue of slipshod taxonomy. First, despite the overlap of some PTSD symptoms with other disorders, there are pathognomonic symptoms of PTSD that make it distinct. No diagnoses other than acute stress disorder (ASD) include trauma-specific items such as criteria B 1–5 (specific to traumatic reexperiencing symptoms) or C 1–3 (specific to traumatic avoidance and numbing). Thus 8 out of 17, or nearly half of the PTSD diagnostic criteria, are unique to PTSD or partly shared by ASD. It is literally impossible to be diagnosed with PTSD without trauma-specific criterion B symptoms. If one conducts a careful interview it should not be difficult to diagnose PTSD even in the presence of other disorders.

Second, Keane and colleagues conducted a study explicitly to address this question. They asked 340 clinicians to rate a 90-item form that was a mixture of items meant to reflect PTSD, MDD, and GAD. MDD and GAD were chosen because they were supposedly two categories with which PTSD was commonly confused. The clinicians were asked to rate the extent to which each item reflected each of the three disorders on Likert scales. The findings were clear that clinicians had little difficulty distinguishing PTSD from MDD or GAD (Keane, Taylor, & Penk, 1997).

Third, most prior studies had at least one of two shortcomings limiting their ability to demonstrate the precise relationship between comorbid disorders and PTSD. The first shortcoming was a failure to ask about the temporal relationships between non-PTSD disorders and traumatic events. The second shortcoming was a failure to consider individuals with any PTSD symptoms as “PTSD cases” so that their post-trauma non-PTSD disorders would appear to have arisen in the absence of PTSD symptoms. For example, if an individual developed MDD and six symptoms of PTSD following a trauma, but the PTSD symptoms were not in the correct algorithm for the diagnosis, this individual would not have been counted as a PTSD case and the new MDD disorder would not have been counted as a comorbid disorder, when in fact they should be. This could have led some observers to assume that the non-PTSD disorders developed after traumas in the absence of PTSD, when plainly this individual had substantial PTSD symptoms. A study by McMillen and colleagues rectified those shortcomings by studying 162 adult flood survivors and carefully tracked the onsets of all symptoms. Of all individuals that developed new non-PTSD disorders following the flood, they all developed either full PTSD or significant PTSD symptoms (McMillen, North, Mosley, & Smith, 2002). A PTSD syndrome seems to be the core post-trauma response. Comorbid disorders may or may not develop in individuals for different reasons. This finding was replicated subsequently with preschool children and with their caregivers following Hurricane Katrina (Scheeringa & Zeanah, 2008).

It is also worth noting that not all of the comorbidity develops following trauma. Findings from studies that examined subjects prospectively prior to exposure to traumatic events showed that a proportion of the comorbid conditions pre-date (and perhaps serve as vulnerability factors for) the development of PTSD. For example, when studied prior to traumatic events, 100% of adults who had PTSD in the last year at age 26 had met criteria for another mental disorder between the ages of 11–21 years (Koenen et al., 2008).

Age Differences in Criteria and Algorithms

Infant and Preschool Children

Ten studies have examined PTSD symptoms with developmentally-sensitive criteria in young children (Bogat, DeJonghe, Levendosky, Davidson, & von Eye, 2006; Ghosh-Ippen, Briscoe-Smith, & Lieberman, 2004; Levendosky, Huth-Bocks, Semel, & Shapiro, 2002; Meiser-Stedman, Smith, Glucksman, Yule, & Dalgleish, 2008; Ohmi et al., 2002; Scheeringa, Peebles, Cook, & Zeanah, 2001; Scheeringa & Zeanah, 2008; Scheeringa, Zeanah, Drell, & Larrieu, 1995; Scheeringa et al., 2003; Stoddard et al., 2006) compared to two such studies in school-age and adolescent children (Meiser-Stedman et al., 2008; Scheeringa, Wright, Hunt, & Zeanah, 2006), so that the field of PTSD is in an unusual place that diagnostic validity has been more meticulously studied in young children than in school-age and adolescent children.

Based on findings in these studies, PTSD can be reliably detected in young children, and an alternative diagnostic algorithm appears more developmentally-sensitive and valid than the DSM-IV criteria. Also, longitudinal studies have indicated that children do not simply “grow out” of stress symptoms and may have a more unremitting course than adults.

One main challenge with infant and preschool children is that symptomatology depends on verbalizations from the individuals making them difficult to detect in preverbal or barely verbal children. For example, detection of the item “avoidance of thoughts, feelings, or conversations associated with the traumatic event,” requires children to verbalize that they experienced these phenomena. This item may be detected somewhat from behavioral observations of avoidance, but this requires an adult’s presence to observe, infer, and report on the child’s behavioral reactions. A second main challenge is that symptoms that are experienced internally are difficult or impossible to detect, even when an adult is present to observe. This holds true even for young children who have developed narrative language capacities. This capacity to self-report on any internalized symptomatology gradually emerges with age, but is not yet a mature skill in young children.

It was clear from reviewing signs and symptoms of highly-symptomatic, traumatized young children that they could not meet the algorithm for the DSM-IV diagnosis of PTSD (Scheeringa et al., 1995). This was largely due to the requirement for three avoidance and numbing (criterion C) items. The seven “avoidance and numbing” items are also the most internal of the 17 possible PTSD criteria. If the requirement for the numbing and avoidance items was lowered to one item, then young children could qualify for the diagnosis at reasonable rates (Scheeringa et al., 2001; Scheeringa et al., 1995; Scheeringa et al., 2003).

The algorithm change is the main change, but changes in wording of five items have also been proposed (Table 1). (1) Do not require the A2 item that the child’s response at the time of the traumatic event involved helplessness, fear, or horror. This transient reaction cannot be known if children are preverbal or adults were not present to witness the children’s reactions.

Table 1.

Proposed Changes to Wording for Five DSM-IV PTSD Items

DSM-IV #	DSM-IV Wording	Proposed Change
A.2	person’s response at the time of the event involved helplessness, fear, or horror.	Add: For young children, make this optional if there was no adult present to witness the children’s reactions.
B.1	recurrent and intrusive distressing recollections of the event	Note: For young children, distress not required.
C.4	markedly diminished interest or participation in significant activities	Note: For young children, significant activities may be play, social interactions, and daily routines.
C.5	feeling of detachment or estrangement from others	Note: For young children, may appear as social withdrawal from loved ones or acquaintances
D.2	irritability or outbursts of anger	Add: or new temper tantrums

The next four modifications are minor wording changes that enhance the face validity of the items for young children, but do not alter the intent of the original DSM-IV items. (2) For the symptom “recurrent and intrusive distressing recollections of the event,” include a note that the distress may not be obvious. Some young children spontaneously talk about or play out their intrusive traumatic recollections but they do not appear obviously distressed. The reason for this is unclear. The initial intuitive interpretation may be to think that children who do not show distress are overall less severe. Yet, there is some evidence from maltreated children that this may reflect a worse prognosis. In a study of adolescent and young adult women who had experienced sexual abuse, those who expressed positive affect when discussing their abuse had poorer long-term social adjustment than those who expressed negative affect (Bonanno et al., 2007).

(3) For the item “markedly diminished interest or participation in significant activities,” it makes relatively more sense to ask about “significant activities” in terms of play, social interactions, and daily routines.

(4) For the item “feeling of detachment or estrangement from others” (a highly internal phenomenon) ought to be asked also in more behavioral terms as social withdrawal from loved ones or acquaintances.

(5) When asking about the item “irritability or outbursts of anger,” it makes reasonable sense to include inquiry about new or extreme temper tantrums.

Using an alternative algorithm and sets of criteria that were similar to the wording changes recommended above, five studies have now calculated rates of PTSD in head-to-head comparisons with the DSM-IV rules (Table 2). The last column of Table 2 also reports the mean number of PTSD symptoms for three of the studies. The frequencies of PTSD in young children identified by using the DSM-IV rules are considerably lower than the frequencies identified by using the alternative algorithm. When coupled with the observation that children diagnosed by the alternative rules were highly symptomatic (6.1 to 10.0 symptoms), this suggests that DSM-IV criteria require some modification for use in young children.

Table 2.

Head-to-head comparisons of DSM-IV and PTSD-AA criteria frequencies of diagnoses in five studies of young children.

	Ages	DSM-IV diagnosis frequency	PTSD-AAa diagnosis frequency	Mean number of PTSD symptoms for PTSD-AA cases
Scheeringa et al., 1995. n=12 (clinic)	1–4 years	13%	69%	Not reported
Scheeringa et al., 2001. n=15 (clinic)	1–3 years	20%	60%	9.9
Ohmi et al., 2002. n=32	2–6 years	0%	25%	6.1
Scheeringa et al., 2003. n=62	1–6 years	0%	26%	6.1
Meiser-Stedman et al., 2008. n=156	2–6 years	1.7%	10%	10.0
Scheeringa et al., 2008. n=70	3–6 years	15.7%	50.0%	7.8

^aScheeringa et al., 1995, Scheeringa et al., 2001, and Ohmi et al., 2002 used an earlier PTSD-AA algorithm that required a total of four symptoms: one criterion B re-experiencing, one criterion C numbing/avoidance, one criterion D increased arousal, and one from a new criterion consisting of three items (new fears not related to the trauma, new separation anxiety, and new aggression. Criterion C included a new item called developmental regressions.

The three later studies used an algorithm that also required a total of four symptoms but in a different algorithm: one from criterion B, one from criterion C, and two from criterion D.

Note that lowering the threshold for criterion C has been suggested even for adults because highly symptomatic and impaired adults could not be given the diagnosis if they had only two but not three criterion C symptoms (Kilpatrick & Resnick, 1993).

One additional note is that research also identified four items of symptomatology that were commonly shown by young children but were not present in the DSM-IV. These four items are (1) regression in skills (e.g., verbal skills, dressing skills, and toileting), (2) new onset separation anxiety, (3) new onset aggression, and (4) new onset of fears of things not related to the trauma (e.g., going to the bathroom alone, and the dark). Including these items in the diagnostic criteria did not increase the diagnostic sensitivity, however (Scheeringa et al., 2003). They still may have other uses, such as constituents of dimensional symptom checklists or treatment outcome measures.

School-Age and Adolescent Children

The two studies that examined diagnostic algorithms in school-age children also found suggestive evidence that the DSM-IV algorithm was not sensitive enough to diagnose children who were highly symptomatic and functionally impaired. A small study of 11 7–11 year-old acute injury victims found a frequency of diagnosis of 18.2% by the alternative 1B/1C/2D algorithm versus 9.1% by DSM-IV criteria (Scheeringa et al., 2006). A study of 52 7–10 year-old motor vehicle accident victims found a frequency of diagnosis of 18.8% by the 1B/1C/2D algorithm versus 2.1% by DSM-IV criteria (Meiser-Stedman et al., 2008); the 1B/1C/2D cases had a mean of 11.6 DSM-IV symptoms (by combined parent and child reports). Again, it seems implausible that children who have so many PTSD symptoms and impairment should not meet the diagnostic threshold for PTSD.

Course

Most individuals who experience truly life-threatening traumas show some symptoms of PTSD within the first month (Davidson, Hughes, Blazer, & George, 1991), but the majority are resilient and do not show enduring symptoms. Approximately 70% of persons lose those symptoms over the first month and only 30% tend to have enduring symptoms (Kessler et al., 1995). This fits fairly well with the DSM-IV requirement that symptoms must be present for one month before a diagnosis is made (keeping in mind, however, that those who eventually qualify for the diagnosis have been symptomatic the entire first month and could have been receiving treatment). Thus, most individuals exposed to life-threatening events do not develop PTSD.

More specifically, longitudinal studies have demonstrated four trajectories of symptoms: 1) a resilient group who develop symptoms but show quick recovery within the first month following severe trauma, or show no symptoms in the first month following less severe traumas; 2) those who develop symptoms but show substantial naturalistic recovery over the first one to two years (symptoms with recovery); 3) those who develop unremitting symptoms (chronic PTSD); and 4) those with delayed onset (Bonanno, Rennicke, & Dekel, 2005).

So these trajectories mean that of those who develop PTSD, it is a chronic illness in approximately 50% of patients (Davidson & Fairbank, 1993). Despite this, the vast majority of longitudinal studies in adults have shown that groups show some significant naturalistic reduction in mean symptoms over time, even if the signs and symptoms do not completely disappear. In children, prospective studies are not yet common, but the early studies suggest a pattern of relatively more unremitting and chronic PTSD. Scheeringa and colleagues studied 62 1–6 year-old children with mixed traumatic experiences and found significant stability of symptoms over two years. The group that had been diagnosed by the 1B/1C/2D algorithm at Time 1 had an estimated mean of 6.1 PTSD symptoms which did not diminish by as much as one symptom over 2 years (Scheeringa, Zeanah, Myers, & Putnam, 2005). Meiser-Stedman et al. (2008), in a prospective design, showed that 69% of preschool children diagnosed with the 1B/1C/2D algorithm at 2–4 weeks post-trauma retained the diagnosis six months later.. A prospective study of 808 Australian schoolchildren (mean 8.2 years) showed no decrease in stress symptoms 26 months after a bushfire (McFarlane, 1987). Seventy percent of 30 6–11 year-old children still were in the moderate to severe category of posttraumatic stress symptoms 21 months after surviving Hurricane Andrew (Shaw, Applegate, & Schorr, 1996).

Unremitting PTSD takes on new salience in early childhood during a time of uniquely rapid brain development. Given the speculations about potentially permanent alterations in the developing brain from early traumas (Perry, Pollard, Blakley, Baker, & Vigilante, 1995), these preliminary data raise concerns about the chronicity of PTSD symptomatology in young children. It would be important to identify the risk factors for chronic PTSD but progress on this issue in children is underdeveloped. At this point, one may only extrapolate from adult studies that have identified in a preliminary way factors that predict chronic PTSD (as opposed to developing PTSD that remits naturally to some degree). Unhappily, just as the development of PTSD so far does not appear to be accounted for by a few risk factors, chronic PTSD also appears to be multifactorial. For example, in a prospective study of 205 assault victims who were assessed with one of the more comprehensive and well-chosen batteries of 13 potential predictor variables, eight of the variables significantly predicted PTSD after six months (Kleim, Ehlers, & Glucksman, 2007). The combination of the best three predictors (mental defeat, rumination about the trauma, and prior psychological problems) accounted for only 47% of the variance.

Parenting and PTSD

Scheeringa and Zeanah (2001) reviewed 17 studies that met the inclusion criteria for their review that (a) children had been exposed to a DSM-IV level trauma, (b) both the children and parents were assessed concurrently, and (c) the studies used measures that were standardized and replicable. All but one of those studies found a significant association between an aspect of poor parental functioning and less adaptive child outcome. The poor child outcomes most often were PTSD symptoms. The parental variables included PTSD, depression, anxiety, overprotectiveness, less family cohesion, family chaos, and perceived less supportiveness (Scheeringa & Zeanah, 2001).

In a study of sexually abused preschool children treated with cognitive behavioral therapy, higher parental emotional distress and depression symptoms at the beginning of treatment predicted those children who improved relatively less (Cohen & Mannarino, 1996). Because of the acknowledged importance of the parent-child relationship for a variety of functions, it is tempting to speculate from these studies that parents who are more symptomatic are less able to protect children who then experience more intense symptomatology.

Yet, a study involving 95 7–12 year-old children, the first to use an observational rating of parenting in relation to domestic violence (DV), showed no simple direction of influence (Levendosky & Graham-Bermann, 2000). After statistically controlling for demographics, past abuse toward the mother, and maternal depression, observed maternal warmth did not predict child outcomes on the Child Behavior Checklist. Observed maternal authority-control was predictive of child behavior, but the author acknowledged that the chicken-or-the-egg question could not be answered by this study design. In other words, children may not have been reacting to their parent’s rearing style, but that the parent’s rearing style might have been a reaction to their children’s characteristics.

In addition, other data with preschool children failed to support the notion that parents influence children following trauma. In the study of 1–6 year-old children mentioned earlier (Scheeringa et al., 2003), parents and children were observed while interacting in structured and unstructured activities, including five minutes of free play, one to two minutes of clean up, and 20 minutes for four structured puzzle tasks (Crowell, Feldman, & Ginsberg, 1988). The final tasks increased in difficulty so that the third and fourth tasks were too difficult for the child to complete alone, thus increasing the demand for dyadic collaboration. Two raters were trained to reliability with an expert rater. Caregivers were rated on eight variables on 1–7 point Likert scales with higher scores meaning better qualities: behavioral responsiveness, emotional responsiveness, positive affect, withdrawn/depressed, irritability/anger, verbal aggression, positive discipline during clean up and negative discipline during clean up.

These previously unpublished observational data were available on 14 children diagnosed by the 1B/1C/2D algorithm with PTSD, 38 with symptoms of PTSD but not enough for the diagnosis (Trauma/No PTSD group), and 61 non-trauma-exposed controls. The caregivers of the PTSD children were significantly more symptomatic for both PTSD and depression. Contrary to expectations, the caregivers of the PTSD children scored highest on emotional responsiveness (Figure 1). In addition, the caregivers of the PTSD showed higher positive affect that was marginally significantly, again contrary to expectations. There were no differences on the other variables. It would be difficult to conclude from these data that negative parenting led to more severe children’s PTSD symptoms.

Figure 1.

Observed parenting qualities of caregivers with children with PTSD (Child PTSD), caregivers with children with PTSD symptoms but not PTSD (Trauma/No PTSD), and caregivers of non-exposed children (Child Control).

Caregivers of the Child PTSD group scored significantly higher Emotional Responsiveness than caregivers of the Child Control group in a Wilcoxon rank-sum pairwise test (p<.05) after a significant three-group Kruskal-Wallis test. Caregivers of the Child PTSD group showed marginally significant higher Positive Affect compared to the other caregivers (Kruskal-Wallis test p=.06).

These are the first observational data of parent-child interaction for children with PTSD symptoms, which provide a different perspective than the 17 self-report studies reviewed by Scheeringa and Zeanah (2001). These findings serve as a gentle caution in drawing premature conclusions about the directionality of effects before the data are in. We actually do not yet know about the direction of effects between children’s and parents’ symptomatology following trauma.

Symptomatic and Impaired, but not Diagnosed

Like any medical or psychiatric diagnosis in which symptomatology exists on a continuum of severity (Angold, Costello, & Farmer, 1999), a PTSD diagnosis dichotomy underestimates the number of children with symptom-related functional impairment. Seven to 14 year-old children with interpersonal trauma who met PTSD diagnostic criteria in only two out of the three criteria B, C, and D had the same level of functional impairment as children who had full PTSD diagnoses (Carrion, Weems, Ray, & Reiss, 2002). When preschool children who suffered heterogeneous types of trauma were re-assessed one year after their first assessment, significantly more were impaired in at least one domain (48.9%) than had the full diagnosis of PTSD (23.4%) (n=47); the gap was even greater after two years (74.3% impaired compared to 22.9% diagnosed; n=35) (Scheeringa et al., 2005).

Current practice parameters recommend that children with clinically significant impairing levels of PTSD symptoms, regardless of diagnostic status, should be provided with evidence-supported treatment options (Cohen & American Academy of Child & Adolescent Psychiatry, 1998). An alternative appropriate diagnosis (e.g. adjustment disorder; anxiety disorder NOS, etc) can be used if PTSD diagnostic criteria are not met. This issue may be reflected in the future DSM-V because it has been suggested for adults to lower the threshold for criterion C from three to two symptoms (Kilpatrick & Resnick, 1993) and for young children from three to one symptom (Scheeringa et al., 2003).

Procedural Validity: Whom to Ask About What?

Children and parental agreements about symptoms are notoriously poor. Each provides some unique information. At least two studies have concurrently assessed the frequencies at which children and their parents report PTSD symptoms. Both studies sampled children who experienced motor vehicle accidents and other acute injuries from emergency departments. In a sample of 24 12–18 year-old adolescents, 8.3% met the threshold for the 1B/1C/2D diagnosis by child report, 4.2% by parent report, and 37.5% by combined report (Scheeringa et al., 2006). In a sample of 51 7–10 year-old children, 17.8% met the 1B/1C/2D diagnosis by child report and 18.8% by parent report, and 40.0% by combined report (Meiser-Stedman et al., 2008).

Given this well-established finding that parents and children each contribute unique information about children’s psychopathology, for which the combination of information is significantly greater than each separately, the ideal assessment of young children would include both informants. Regrettably, interviews of the children themselves when younger than seven years of age are not feasible because they have not yet mastered multiple types of skills needed for this task. Despite some advances in this area with five- and six-years-old children, most notably with the Berkeley Puppet Interview (Measelle, Ablow, Cowan, & Cowan, 1998), there is little reason to believe that children younger than five years would have sufficient skills, and there have been no known studies with children younger than seven years on their accuracy to self-report in relation to diagnoses. Assessments of disorders in young children with current techniques are therefore practically dependent on interviews of their caregivers.

An attempt was made to systematically elicit PTSD symptom information from 1–3 year-old children in a study with structured and unstructured interaction formats (Scheeringa et al., 2001). The interactions were designed to be those that could be realistically used in a clinic assessment setting (e.g., caregiver-child free play, or examiner-guided trauma reenactment play). Only 12% of the children’s symptoms that were reported from parental interviews could be detected from observations and interactions with them.

Even when relying on caregivers’ reports, there is cause for concern that interviews about young children may be less reliable and/or valid compared to older age groups for which there are more established norms for problem behaviors. Nevertheless, when examined empirically, the first reported psychometrics for an early childhood diagnostic instrument showed promising results. The inter-rater reliability of two clinicians rating videotapes of another clinician’s interviews of 15 parents of 1–3 year-old children using a standardized instrument for posttraumatic stress disorder (PTSD) was substantial with a Cohen’s kappa of 0.74 (Scheeringa et al., 2001). Subsequently, the largest demonstration of feasibility of interviewing caregivers of young children comes from a study with the Preschool Age Psychiatric Assessment (PAPA), which was the first multi-disorder instrument with published psychometric properties. The PAPA was used to interview caregivers of 307 2–5 year-old children recruited from a general pediatric clinic by two research assistants for 12 disorders (Egger et al., 2006). Categorical agreement for PTSD was good with a Cohen’s kappa of .73 and continuous agreement was good with an intraclass correlation coefficient of .56.

Interviewing Method

The implications from the above for interviewing respondents are organized below into seven practical suggestions.

1. Professionals must be aware that young children can develop PTSD. Only then can appropriate screening and referrals for assessment be triggered.

2. Even with older children, self report of symptoms is rarely spontaneous—only systematic inquiry will successfully elicit accurate data.

3. When conducting assessments, developmentally-appropriate measures and criteria must be used so as not to miss the diagnosis.

4. When new-onset comorbid disorders arise following trauma, typically ODD and SAD in young children, this ought to trigger a full assessment for PTSD.

5. While the DSM-IV criteria do not restrict making the diagnosis to a single traumatic event, standardized instruments for PTSD often ask respondents to select “the worst” traumatic event that he or she experienced and to rate all PTSD symptoms in relation to that specific event. Yet, one recent study indicated that 37.0% of all children in the U.S. have experienced more than one potentially traumatic event (PTE) (Copeland et al 2007). It is often difficult for children, particularly young children, to select only one traumatic event as “the worst” they have experienced. Because it would be a bit absurd clinically to restrict one’s attention to only a portion of children’s PTSD symptoms, it seems logical that an instrument be comprehensive and assess all PTSD symptoms from all past events.

6. In order to minimize both false positives and false negatives, one must conduct a comprehensive, standardized, and rigorous interview for all 17 symptoms of PTSD. Specifically, one must ask from a menu of probes, and ask for examples, onsets, durations, and frequencies. This means using a structured instrument.

7. Most importantly, clinicians must probe further than asking “does your child have x symptom?” A simple yes or no response can mean almost anything. One has to obtain examples. If the respondent cannot provide examples, the interviewer must provide possible examples. This type of educational interviewing gives respondents a frame of reference for the internalized and abstract items comprising signs of PTSD. This is in contrast to other types of symptomatology, such as hyperactivity or depression, which are readily observable and intuitively obvious to most people. PTSD symptoms are not intuitively obvious or experienced by most people. Concerns that this type of interviewing is inappropriately “leading the witness” who will endorse symptoms that do not exist are typically unjustified because those types of concerns are for malingerers, and there is the failsafe that the gathering of examples will always be able to support or disprove a respondent’s answer.

Treatment Implications

Evidence-based treatments now exist for children with PTSD for nearly all ages. The major issue is not whether they work but whether children who need them receive access to these treatments as they are undoubtedly underutilized. For example, rapid needs assessments conducted immediately after the World Trade Center disaster (Herman, Felton, & Susser, 2002) and the Hurricane Katrina disaster (Dalton, Scheeringa, & Zeanah, 2008) estimated that 520,000 and 260,612 individuals respectively would develop PTSD. Treatment capacity could never treat that many children, but even if it could, the obstacles noted earlier to disorder recognition and assessment challenges would prevent many thousands from getting treatment.

Despite the well-known phenomenon of comorbid disorders with PTSD from all types of traumatic events (Scheeringa et al., 2003) and claims of cases receiving inappropriate treatment for their comorbid disorders (van der Kolk, 2005), there is little evidence that comorbid disorders themselves cause incorrect treatment decisions. Instead, it is more likely that if inappropriate treatment decisions are being made, it is because clinicians are not appropriately assessing for PTSD. If confusion exists from the presence of comorbidity, it is not inherently a flaw of the taxonomy system in general or PTSD specifically. Good history taking about the timing of symptom onset and knowledge of the research that PTSD is the underlying basis of new disorders after trauma exposure should contribute substantially to accurate diagnosis and treatment planning.

There are few data on whether having a second (or third) disorder makes it more difficult to treat PTSD, or vice versa, whether having PTSD makes it more difficult to treat a co-existing disorder. It would be intuitive to hypothesize that a cumulative load of disorders predicts greater chronicity and greater treatment resistance. This is an important area of research that may be able to identify subsets of children with different courses and different biomarker profiles who require alternative treatment modalities to achieve equivalent effect sizes.

While the brief data presented earlier on parenting issues suggested an important direction for assessing parent and family issues, and attempting to understand more clearly the impact of those issues on child behavior, caution is warranted in attributing directionality. There is in fact no evidence that children cannot improve when treated with evidence-based therapy regardless of degree of parental symptomatology or family dynamics. The number of controlled studies investigating the addition of parents to children’s treatment is extremely limited and these do not generally support adding parental components to therapy (Deblinger, Steer, & Lippmann, 1999). Even in the Cohen et al study (1996) which suggested that young children improved less who had parents who were more symptomatic at the beginning of therapy, the children who had the more symptomatic parents still improved with therapy. It would be premature, for example, to withhold or delay treatment for children while parents are referred to their own treatment, but more research is evidently needed in this and other interesting areas.