Prospective measurement of skin conductance response during trauma interview predicts future PTSD severity in trauma exposed children

Abstract

Previous cross-sectional studies have shown that sympathetic nervous system (SNS) arousal is positively associated with posttraumatic stress disorder (PTSD) symptoms in children with trauma exposure. One of the ways that SNS activity is measured is through skin conductance response (SCR), which has been shown to predict future PTSD severity in adults. In this study, we explored the utility of a novel, low-cost mobile SCR device, eSense, to predict future PTSD symptom severity in trauma exposed children. We recruited children (N = 43, age 9 years at initial visit) for a longitudinal study in which SCR was recorded at baseline visit, and PTSD symptoms were assessed two years later. Results indicated an interaction between SCR and trauma exposure, such that children with lower trauma exposure who demonstrated greater SCR reported higher PTSD severity two years later. This association remained significant even after controlling for baseline PTSD symptoms. Children with higher levels of trauma exposure did not show this association, potentially due to ceiling effects of PTSD symptoms. Together these findings suggest the utility of SCR as a biomarker for predicting trauma related disorders in children, and that it may be a valuable tool in clinical interventions targeting sympathetic arousal.

1. Introduction

Trauma refers to an event that poses an actual or perceived threat to an individual’s life, such as physical or sexual assault, a motor vehicle accident, or a life-threatening illness [1]. In the United States approximately 60% of adolescents have experienced at least one potentially traumatic event [2]. Traumatic events can be directly experienced, witnessed, or learned about from others. While about 20% of youth experience trauma, only 5% meet diagnostic criteria of post-traumatic stress disorder (PTSD) [3]. This is because while some children may be significantly affected by traumatic events, they may not meet the full criteria for PTSD at an early age. The Diagnostic and Statistical Manual of Mental Disorders (DSM-5) defines PTSD based on five criteria, including experiencing a traumatic event, intrusive thoughts about the trauma, avoidance, negative mood and cognition, and hyperarousal symptoms. Many children who have experienced trauma exhibit sub-threshold PTSD symptoms but still face functional impairment in their daily lives, and may benefit from treatment [4].

Children with trauma exposure may have an increased vulnerability to develop PTSD in the future due to genetic, physiological, or neural factors. Therefore, identifying biomarkers of risk for psychopathology is critical for early intervention or even prevention of PTSD symptoms. Current assessments of PTSD severity rely primarily on self-report. However, children's subjective reports may not always accurately reflect the severity of their symptoms. Therefore, incorporating physiological measures alongside trauma interviews may assist treatment clinicians in identifying at-risk individuals who might otherwise go untreated. One such measure that has shown promise in relating self-report of PTSD symptoms with physiological arousal is skin conductance response (SCR) [5].

Skin conductance response is a peripheral measure of sympathetic nervous system (SNS) activity and is triggered by arousing stimuli, such as affective images or memories related to past traumatic events. SCR increases in response to both positively and negatively valanced stimuli and is associated with sudden arousal and anxiety related to salient traumatic memories [6]. Previous studies in adults with PTSD have demonstrated SCR measured during a trauma interview can predict current [7], as well as future [8] PTSD symptoms severity. Although research on skin conductance in children is limited, a few studies have explored changes in skin conductance during fear conditioning paradigms, where children learn to associate a threat cue with an aversive outcome. Children who have experienced maltreatment often exhibit blunted skin conductance responses to threat stimuli during fear conditioning [9]. A recent study from our lab[10] revealed that children with higher trauma exposure demonstrated higher SCR during a trauma interview. Another study conducted with resettled refugee youth in the United States also found a positive correlation between SCR during a trauma interview and the number of traumatic events experienced by the youth [11]. These findings suggest that SCR may reflect an underlying state of heightened autonomic arousal associated with an increased risk of trauma-related psychopathology. However, the studies above were cross-sectional and were not able to predict future risk. In adults with trauma exposure, mobile measurement of heightened SCR in the emergency room hours after trauma predicted PTSD symptom severity 6 months later [8]. The current study applied this same method to assessing future PTSD severity in a longitudinal study of school-age children with trauma exposure.

There is evidence indicating that heightened SNS activity could increase risk for trauma-related psychopathology during development. Early-life stress may disrupt the sympathetic adrenal-medullary (SAM) system, resulting in rapid SNS activation and increased release of catecholamines from the adrenal medulla [12]. For example, a study of girls who experienced sexual abuse found elevated urinary catecholamines [13]. Catecholamines including norepinephrine subsequently trigger a cascade of physiological reactions including increased heart rate, accelerated breathing, and heightened SCR. Consequently, higher skin conductance levels may indicate higher norepinephrine levels, reflecting an abnormal stress response that can lead to increased PTSD symptoms. Salivary alpha amylase, an enzyme in saliva associated with stress-induced SNS activity [14], has also been found to be positively correlated with PTSD symptoms in adolescent girls who experienced sexual abuse [15]. SNS activity is regulated in part by limbic structures of the brain, including the amygdala and hippocampus [16]. Previous studies of youth with trauma exposure have indicated that the amygdala is overactive in individuals with higher anxiety symptoms [17]. On the other hand, age-related functional connectivity between the hippocampus and insula interacts with level of trauma exposure, such that children with lower violence exposure showed decreased connectivity with age [18]. Measuring SNS activity using SCR during an arousing event, such as a trauma interview, may enable clinicians and researchers to assess the magnitude of this rapid stress response and the risk of developing psychopathology, particularly PTSD.

Considering the existing studies that have found a significant association between skin conductance and PTSD symptoms in children [10], [11], our study aimed to investigate whether skin conductance assessed during a trauma interview at nine years of age would predict severity of PTSD in children two years later. We hypothesized that SCR measured during the initial visit would demonstrate a positive correlation with PTSD symptoms observed at the follow-up visit. In addition, we tested whether the degree of exposure to violent events would moderate this association.

2. Material and methods

2.1. Participants

Participants were recruited from the metropolitan Detroit area and enrolled in a longitudinal study. Inclusion criteria for the current study were: age 9 years at the initial visit and willingness to participate with a legal guardian. Exclusion criteria were: autism spectrum disorder, psychotic disorder, or cognitive disability. All caregivers provided informed consent for both them and their child. The child provided verbal assent. All study procedures were approved by the Wayne State University Institutional Review Board prior to recruitment. Data from the children’s initial visit have been published previously [10], however this paper includes the longitudinal data from the follow-up visit two years later. The current sample included N = 43 children with SCR data from the initial visit and with PTSD symptom data from the follow-up visit. More than half of the children (28/43; 65%) identified as female and 31/43 (72%) of the children identified their race as Black, 8/43 (19%) identified as White, and 4/43 (9%) identified as Other. Most of the participants identified as non-Hispanic (40/43; 93%). See CONSORT diagram in Fig. 1 for participant flow.

Fig. 1.

CONSORT diagram of participant flow.

2.2. Skin conductance measurement

During the initial visit, all children underwent a trauma interview while we simultaneously measured skin conductance using eSense (Mindfield Biosystems, Inc.), an app that records skin conductance level (SCL) using a iPad tablet. Prior to the interview, isotonic electrode paste and electrodes were applied to the child's non-dominant index and middle fingers. These electrodes were then connected to an iPad via a headphone jack cable. Skin conductance data using the eSense device was collected at a sampling rate of 5 Hz. Consistent with previous studies, a 2-minute baseline was recorded to establish the child's resting skin conductance level while they sat quietly and remained calm. Following the baseline, the eSense recording was paused, and the child was provided with instructions regarding the trauma interview. eSense was then restarted and skin conductance was continuously recorded during the interview. This method of SCL data collection has been externally validated against gold standard psychophysiological recording equipment in our prior research [7]. In the current study we tested the internal consistency of the measure by comparing the different phases of SCL recording. We found excellent internal consistency, as baseline SCL was highly correlated with SCL at the start of the of the trauma interview, r = 0.94, p < 0.001. As in previous research [19], [7], we quantified the children's SCR by calculating the maximum SCL during the interview and subtracting the last 30 s of the baseline SCL from it. Additionally, we assessed baseline SCL prior to the interview, and SCL habituation by subtracting the SCL at the end of the interview from the maximum SCL.

2.3. Clinical assessments

The number of criterion A traumatic events was assessed using the Trauma Events Screening Inventory (TESI), a verbally administered questionnaire designed to assess the number of traumatic events a child has experienced or witnessed in their lifetime [20]. The interview follows a structured format that encourages the child to actively recall their memories associated with trauma. The interview includes questions about various traumatic events such as domestic and community violence, injuries, natural disasters, verbal and sexual abuse. If a child acknowledged experiencing any of the queried traumas, they were further prompted for specific details, including the nature of the event, their age at the time, and the individuals involved. In this study, we focused on analyzing the total number of reported traumas, encompassing both personally experienced traumas and those witnessed by the child. The TESI has both a parent and child report; for this study we focused on child report of trauma, as that was the measure associated with SCR in our prior study [10].

The Violence Exposure Scale Revised (VEX-R) is a comprehensive trauma interview consisting of 30 questions, specifically designed to assess the number and severity of violent events experienced by children [21]. This interview has been validated in pediatric urban trauma populations [22] and uses a visual approach, presenting a series of pictures depicting characters witnessing or directly experiencing violent events. Following each picture, the child was queried about the frequency with which they have encountered or witnessed a comparable situation to the one portrayed in the image. At the end of the interview, the number of endorsed violent events were summed.

PTSD symptoms were assessed at the initial visit and during the follow-up assessment approximately two years after their initial visit. We administered the University of California Los Angeles Posttraumatic Stress Disorder Reaction Index for DSM-5 (UCLA-RI-5) to assess trauma-related PTSD symptoms in children [23]. The UCLA-RI is a well-validated assessment consisting of 27 items that assess the severity of PTSD symptoms experienced by the child within the last month. Both child self-report and parent report forms are available. Excellent internal reliability and test-retest reliability have been reported for the DSM-IV, and considerable data is available regarding the instrument’s validity [24]. The UCLA-RI-5 has been updated and validated for the DSM-5, with the child self-report version showing excellent internal consistency in ages 7–18 (Cronbach ɑ=0.94–0.96), as well as high diagnostic sensitivity (1.00) and specificity (0.86) with a score cut-off of 35 [23]. In the current study, the UCLA-RI was administered to both children and their parents, separately (see Table 1 for total scores). Notably, child-reported exposure to violence (VEX-R) and trauma (TESI) were associated with child-reported PTSD symptoms (r = 0.46 and r = 0.43, p’s < 0.005, respectively), but not with parent-reported PTSD symptoms in their child (r = −0.012 and r = −0.02). Given the excellent reliability of child-report and its strong association with trauma exposure, in the current study we used child-reported PTSD symptoms as the primary outcome.

Table 1.

Participant characteristics in the total sample and in both trauma groups. Trauma groups were based on violence exposure and categorized using a median split of the self-report VEX-R measure. *p < 0.05, * *p < 0.01. Acronyms: Violence Exposure Scale Revised (VEX-R); Trauma Events Screening Inventory (TESI); University of California Los Angeles Posttraumatic Stress Disorder Reaction Index for DSM-5 (UCLA-RI-5); Mini International Neuropsychiatric Interview for Children and Adolescents (MINI-Kid).

	Total Sample N = 43	Lower Trauma N = 21	Higher Trauma N = 22
Child Sex/Gender
Male	17 (40%)	7 (33%)	10 (45%)
Female	26 (60%)	14 (67%)	12 (55%)
Child Race/Ethnicity
Black/African American	31 (72%)	17 (81%)	14 (64%)
White	8 (19%)	3 (14%)	5 (23%)
Other	4 (9%)	1 (5%)	3 (14%)
Parental Education
Highschool or GED	25 (58%)	14 (67%)	11 (50%)
Some College	8 (18%)	3 (14%)	5 (23%)
College Degree or above	10 (23%)	6 (29%)	4 (18%)
Trauma Report Mean (SD)
Child Reported Trauma Exposure (TESI)	4.64 (2.52)	3.43 (2.06)	5.86 (2.37)* *
Parent Reported Trauma Exposure (TESI)	5.37 (3.32)	5.43 (3.70)	5.32(2.99)
PTSD Symptoms Mean (SD)
Child Reported PTSD Sys (UCLA PTSD-RI)	26.86 (13.75)	22.78 (11.94)	30.74 (14.51)*
Parent Reported PTSD Sys (UCLA PTSD-RI)	19.91 (12.72)	21.00 (12.96)	18.94 (12.81)
Neuropsychiatric Assessment (MINI-Kid)
Social Anxiety Disorder (SAD)	3 (7%)	2 (10%)	1 (5%)
Obsessive Compulsive Disorder (OCD)	3 (7%)	1 (5%)	2 (9%)
Generalized Anxiety Disorder (GAD)	1 (2%)	0	1 (5%)
Attention Deficit Hyperactivity Disorder (ADHD)	3 (7%)	0	3 (7%)

In order to include a broader range of diagnoses and to examine comorbidity with other DSM disorders, we conducted a child interview using the Mini International Neuropsychiatric Interview for Children and Adolescents (MINI-Kid [25]). The MINI-Kid is a semi-structured diagnostic interview designed to assess current and past episodes of psychopathology in school age children and adolescents (ages 6–18) according to DSM-5 criteria. Comorbid disorders are listed in Table 1.

2.4. Data analyses

Repeated measures ANOVA (RMANOVA) were used to test longitudinal effects on PTSD symptoms with time (2 levels: initial visit, follow-up visit) as within-subjects factor, and trauma group (2 levels: higher, lower) as a between-groups factor. As in our prior work, the VEX-R was used to categorize children into two groups based on a median split of exposure to violence events: a higher trauma group and a lower trauma group. In order to test whether the SCL increased during the trauma interview relative to baseline, and whether this increase is moderated by degree of trauma exposure, we conducted a 2 × 2 RMANOVA with SCL (2 levels: baseline, trauma) as the within-subjects factor and trauma groups as the between-groups factor. SCR was derived as a change score between the trauma interview and baseline; this variable was used as a predictor of PTSD symptoms at follow-up. We conducted a hierarchical stepwise regression analysis predicting future PTSD symptoms by entering the level of violence exposure and PTSD symptom severity at the baseline visit, as well as sex, as the first step, SCR as the second step, and VEX-R by SCR interaction as the third step. The covariates were selected based on prior literature showing that level of trauma exposure, prior PTSD symptoms, and female sex are strong predictors of PTSD symptoms, and we aimed to test whether SCR explained variance in symptoms above and beyond these factors. In order to follow up significant interactions and test for differential patterns in the children with higher and lower violence exposure, we performed Pearson correlations and regressions between SCR and future PTSD symptoms separately in each group.

3. Results

3.1. Participant characteristics

Table 1 shows demographic and clinical characteristics of the participants stratified by violence exposure on the VEX-R. There were no group differences in age, sex, race/ethnicity, or parental education, however, child-reported PTSD symptoms were greater (M=30.48, SD=13.80, range 3–59) in the group with higher VEX-R scores compared to those with lower VEX-R scores (M=20.27, SD=12.51, range 0–42), F(1,42)= 6.46, p = 0.015. Parent-reported PTSD symptoms did not differ between groups. While the average of the UCLA-RI-5 for the entire sample (M=25.26) was below the diagnostic cutoff of 35, 10 of the 43 children (23%) met threshold for PTSD diagnosis at the initial visit. Table 1 also includes comorbid diagnoses from the MINI-Kid.

3.2. Trauma exposure and PTSD symptoms

A 2 × 2 RMANOVA of PTSD symptoms at each time point with time (2 levels: initial visit, follow-up visit) and trauma group (2 levels) showed no main or interaction effect of time, but a main effect of group, F(1,38)= 4.29, p < 0.05. Neither group showed a significant change in PTSD symptoms between the initial and future visit, as shown in Fig. 2.

Fig. 2.

The higher trauma group had greater PTSD symptoms than the lower trauma group across both visits; main effect of group, F= 4.29, p < 0.05.

3.3. Skin conductance during baseline and trauma interview

Skin conductance levels were measured during the initial visit while asking children about their traumatic experiences on the TESI. Fig. 3 shows the SCL during baseline and trauma interview across both trauma groups. The results showed a robust increase in SCL from baseline to trauma interview, F(1, 41)= 71.17, p < 0.001. However, there were no main or interaction effects of trauma group (Fig. 3). Therefore, our findings show that SCR, as measured by the eSense mobile device, consistently increased during the trauma interview regardless of the severity of trauma.

Fig. 3.

Skin conductance levels increased significantly during trauma interview relative to baseline in both trauma groups; main effect of interview, F= 80.72, p < 0.001.

3.4. SCR and future PTSD symptoms

A regression analysis was used to test whether SCR at the initial visit predicted future PTSD symptoms, while controlling for baseline PTSD, sex, and baseline trauma exposure. In a hierarchical stepwise regression, these covariates were entered as the first step and SCR and VEX-R (continuous score) were entered as a second step, and the SCR*VEX-R interaction term was entered as the third step. The final model was significant, F(6, 36)= 4.15, R² = 0.45, adjusted R² = 0.34, p = 0.004. There was a main effect of VEX-R, β = 0.68, t = 2.34, p = 0.03, as well as a main effect of SCR, β = 1.09, t = 2.85, p = 0.008, on future PTSD symptoms. The interaction between SCR and VEX-R was also significant, β = −1.85, t = −2.61, p = 0.01.

We then tested associations between SCR at the initial study visit and PTSD symptoms at the follow-up visit, stratified by violence exposure group. Fig. 4 shows scatter plots of associations within each trauma group. We found a significant positive correlation between SCR and future PTSD severity reported by children on the UCLA-RI-5 (r = 0.49, p < 0.05) among individuals in the lower trauma group. In the higher trauma group, there was a non-significant negative correlation between SCR and future PTSD symptoms, r = −0.20. A regression analysis was then used to test associations beyond the baseline level of PTSD symptoms and trauma exposure in the lower VEX-R group. SCR accounted for a significant change in variance above and beyond baseline trauma and PTSD, R² change= 0.223, p = 0.045, R² = 0.358, adjusted R² = 0.220. In the higher trauma group, the trauma and PTSD symptoms at baseline explained significant variance in future PTSD symptoms, R² = 0.432, adjusted R² = 0.360, p = 0.011; however, SCR did not significantly account for additional variance, R² change= 0.057.

Fig. 4.

SCR was associated with higher severity of future PTSD symptoms in children with lower trauma r = 0.49, p < 0.05.

4. Discussion

PTSD is a disorder that can develop as a result of childhood trauma exposure—pediatric PTSD can emerge in childhood and persist well into adulthood [26]. However, longitudinal data on pediatric PTSD are still sparse, and it is unclear why some children recover from traumatic events while others do not. The primary goal of this study was to explore the clinical and diagnostic utility of a mobile and low-cost device that measures sympathetic arousal through skin conductance, known as eSense, in prospectively predicting future PTSD severity among children with a history of trauma exposure. Our hypothesis was that elevated SCR, an index of heightened SNS activation, would be associated with higher future PTSD symptoms. Our findings supported this hypothesis while also revealing distinctions in skin conductance dynamics between individuals with varying degrees of trauma exposure. We found that SCR associations with future PTSD symptoms differed between individuals with lower and higher trauma exposure, such that the lower trauma group showed a positive association with PTSD symptoms, while the higher trauma group did not. The SCR in individuals with lower trauma exposure was associated with child report, but not caregiver report, of future PTSD symptoms, suggesting self-report of trauma may provide more accurate depiction of underlying sympathetic arousal. This is consistent with previous studies that have found stronger association between child reported trauma and psychopathology in children [10], [17].

Furthermore, our study contributes to the growing literature supporting the use of mobile measurement of skin conductance (SC) in traumatized populations [7], [10], [11] showing that trauma-related questionnaires, when compared to baseline measurements, consistently increase SC. This suggests that skin conductance can reliably be used as a psychophysiological marker in the context of trauma assessment in a variety of samples. This is the first study to use this measure in a longitudinal developmental study to prospectively predict future PTSD symptoms in children.

Our findings on the association between SCR and PTSD severity show heterogeneity, differentiating individuals between higher and lower trauma-exposure groups. We expected that individuals in the higher trauma group would exhibit a robust positive correlation between SCR and future severity of PTSD symptoms. However, we found this association to be true only in the children with lower levels of trauma exposure. Importantly, this association in the lower trauma group remained significant after covarying for degree of trauma and baseline PTSD symptoms. In the higher trauma group, baseline trauma and PTSD were the strongest predictors of future PTSD severity suggesting that this group may have ceiling effects that mask any association with SCR. This notion is supported by the finding that both lower and higher trauma groups showed similar SCR to the trauma interview.

Our findings extend and add to previous cross-sectional research, which has established a positive association between SCR and level of trauma exposure [10]. However, our study also highlights that SCR may exhibit a closer relationship with future PTSD symptom severity, one potentially moderated by the level of trauma experienced. Given that our study included children who were nine years old at the initial visit and then followed up on average two years later, the results may point to early markers of risk for emerging PTSD symptoms. Although the lower trauma group may have been at lower risk at the start of the study, there are likely individual differences in vulnerability that may be related to sympathetic arousal [8]. For instance, individuals in the lower trauma group may have more salient traumatic memories, which could lead to heightened skin conductance responses during trauma-related interviews. Although skin conductance may be a good proxy of hyperarousal symptoms, it’s utility increases when used with other neurophysiological measures. For instance, other proxy biomarkers of SNS activity, such as salivary alpha amylase, have also been associated with increased PTSD severity [15] and hyperarousal symptoms related to stress related psychopathology [27]. A recent theoretical model of pediatric PTSD suggests that greater connectivity within the salience network may be a risk factor for trauma-related psychopathology [26]. The salience network, which includes amygdala and insular cortex, has been found to moderate the association between SCR and subjective arousal [28]. In addition, the cingulate cortex may directly regulate SNS activity such as heart rate (HR) and skin conductance, as deep brain stimulation of the cingulate produces increases in HR and SC via white matter tracts [29]. Measuring SCR provides a low-cost, peripheral marker of these neurobiological underpinnings.

It is important to note that the children in the lower trauma group were not without trauma, endorsing experiencing at least 3, and on average 6.5 violent events. While children in the higher trauma group may benefit from early intervention such as PTSD treatment, the children in the lower trauma group with higher SCR may benefit from prevention strategies. For example, meditation may strengthen anterior cingulate cortex regulation of sympathetic activity in high-risk individuals [30]. Given that heightened SNS activity, including increased norepinephrine release may be a risk factor for developing PTSD in youth, several pharmacological studies have reported positive effects of anti-adrenergic agents, such as propranolol, prazosin, as well as alpha-2 agonists, such as clonidine and guanfacine [31]. While these medications show some promise and appear to be well-tolerated, most of the data rely on open-label and case studies, and more definitive clinical trials are needed [31]. Future randomized clinical trials could stratify by trauma exposure or SCR in order to personalize and optimize such treatments. In the current growing child and adolescent mental health crisis, such targeted strategies are of critical importance [32].

This study had several limitations. For example, this study included a narrow age range at baseline, i.e., 9 year old children, and it is plausible that the physiological changes resulting from traumatic experiences may not have fully manifested to enable the capture of pertinent trauma-related physiological changes. Further, although the follow-up visit occurred two years later, most of the children were still in early adolescence at this time point, and some symptoms may not yet have started to emerge [33]. Consequently, conducting longitudinal analyses of SCR in older children may offer a better understanding of the association between SCR, trauma exposure, and potential future psychopathology. On the other hand, the narrow age range also reduced the impact of developmental changes with respect to brain development, trauma exposure, and physiological changes such as puberty. The mean self-reported PTSD severity score on the UCLA-RI-5 was 25.26 at baseline which is below the clinical threshold of 35. Hence, future studies should investigate SCR in populations with higher PTSD scores. On the other hand, our study found the strongest associations with future symptoms in the individuals with lower trauma suggesting that this measure may be sensitive to detecting at-risk children rather than those already exhibiting severe symptoms. Future studies should also be completed on larger samples with older children and adolescents to determine consistency of the findings. Moreover, neuroimaging studies should assess whether these SCR variations during trauma interviews are modulated by neurobiological changes within corticolimbic structures.

5. Conclusions

This is the first study to use a longitudinal design in school-age children with trauma exposure to examine whether SCR prospectively predicts future PTSD symptoms. We found that heightened SCR during a trauma interview may be a risk factor for increased PTSD symptoms in early adolescence. In the current youth mental health crisis, early biomarkers of risk could lead to targeted prevention strategies for at-risk individuals.

Funding

This work was supported by the National Institutes of Health (R01 MH111682; R01 MH111682-03S1).

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.