Autonomic Nervous System Correlates of Posttraumatic Stress Symptoms in Youth: Meta-Analysis and Qualitative Review

Introduction

Exposure to adverse childhood experiences (ACEs), including exposure to traumatic events, can have damaging effects on biological systems, increases the risk for development of physical and mental health problems, and contributes to substantial health costs across the lifespan (Felitti et al., 1998; National Scientific Council on the Developing Child, 2020). While exposure to ACEs is a nonspecific risk factor for psychopathology, one primary outcome of exposure to ACEs and traumatic events is posttraumatic stress disorder (PTSD) and posttraumatic stress symptoms (PTSS) (Curran et al., 2018). The diagnostic criteria for PTSD include exposure to several types of ACEs, including actual or threatened death, serious injury or sexual violence (American Psychiatric Association, 2013). Another core diagnostic criterion of PTSD is marked alterations in arousal and reactivity associated with the traumatic event to which the individual was exposed. While it is generally accepted that heightened arousal and reactivity is associated with PTSD, there is substantial variability in findings (Zoladz & Diamond, 2013). This paper reviews the literature on the relevance of the autonomic nervous system (ANS), consequences of trauma, and the relation between the ANS and PTSD, quantitatively assesses the relation between multiple physiological measures and PTSS in youth, and qualitatively reviews important additional considerations for future research.

Significance of the Autonomic Nervous System

The ANS serves to regulate a wide range of the body’s involuntary functions, including blood pressure (BP), thermoregulation, gastrointestinal function, pupillary response, and sexual function, as well as responses required to respond to threat (Gibbons, 2019; Kemeny, 2003). The ANS is comprised of two branches: the sympathetic nervous system (SNS) and parasympathetic nervous system (PNS) that have a multitude of opposing and complementary functions. The SNS governs the body’s “flight or fight” response, and controls the mobilization of physiological resources to prepare for physical activity in response to environmental challenges (McCorry, 2007). The PNS governs the “rest and digest” and “feed and breed” responses which return the body to a rest state and facilitate growth and restoration (McCorry, 2007). While coordination of the SNS and PNS facilitates flexibility in meeting environmental demands, a common misconception is that SNS activation causes PNS activation or inactivation, when in fact the neurotransmitters modulating each systems differ (i.e., SNS via norepinephrine and PNS via acetylcholine) and many physiological measures represent coactivation in target organs served by both systems (Levenson, 2014). Therefore, assessment at the subsystem level (e.g., PNS, SNS) can reveal important differences in physiological responding.

ANS involvement in stress responses.

Polyvagal theory emphasizes that physiological states determine a range of social behaviors and emotion regulation abilities (Porges, 2007). The ANS plays an important role in the experience of emotions and responses to environmental stressors. The development of the ANS can provide insight into how and when certain responses to stress may manifest (Porges, 2003b). The PNS-mediated myelinated vagus is integrated into the social engagement system (i.e., social expression, receptivity, and bonds), and when social engagement is compromised, the ANS shifts to activate adaptive defensive behaviors (Porges, 2003a, 2007). This results in the removal of regulatory influences of the myelinated vagus on the heart, which then triggers two older systems: the SNS “fight or flight” and the PNS unmyelinated vagus-mediated “freeze” response (i.e., death feigning). The adaptability of social engagement or defense responses depends on the environment (Porges, 2007). Differences in system activation could result in either increased (i.e., mobilization) or decreased (i.e., immobilization) behaviors and highlights the range of potential physiological and behavioral responses to stress.

Vagal tone is one index of tonic PNS control over heart rate (HR) via the vagus nerve and reflects inhibitory influences that produce a lowered HR compared with the basal firing rate (Porges, 1995, 2007). Vagal tone can be measured noninvasively via respiratory sinus arrythmia (RSA). RSA reflects the rhythmic fluctuation of HR, and is quantified as the degree to which HR increases with inspiration and decreases with expiration (Hinnant et al., 2017). Increased vagal influence on the heart is thought to reflect effective social engagement and perception of safety (Porges, 2011). Higher RSA at rest is conceptualized as a marker of flexible and context-appropriate modulation of ANS activity, and during challenge or threat, the vagal control can be withdrawn, leading to increased HR and metabolic output (Beauchaine, 2001). Therefore, RSA withdrawal (i.e., reduced RSA) during challenge reflects flexible adaptation to environmental demands, and is important for children’s self-regulation, emotion regulation, cognitive control, and adaptive functioning (Beauchaine & Thayer, 2015).

Pre-ejection period (PEP) is a noninvasive index of SNS control over the heart (Newlin & Levenson, 1979). PEP is reflected by the period of time between the electrical initiation of the heart beat and the time that blood is ejected into the aorta (Berntson et al., 1994). In youth and adults, PEP has been shown to increase in response to stress (Berntson, et al., 1994; Quigley & Stifter, 2006). Of note, various factors (e.g., cardiac abnormalities, drug effects) can alter the interpretation of PEP (Newlin & Levenson, 1979). While RSA and PEP are reasonable estimates of parasympathetic and sympathetic control over the heart, respectively, it is important to note that neither is a pure estimate of PNS or SNS function, as HR fluctuations are under joint control of the two systems (Grossman & Taylor, 2007).

Skin conductance is a common SNS measure indexing electrodermal activity (EDA) generated by sweat glands. EDA includes both baseline tonic skin conductance level (SCL) and rapid phasic components that reflect skin conductance responses (SCR) resulting from sympathetic neuronal activity. In contrast to other sympathetic indices (e.g., PEP), SCL is a useful index of sympathetic arousal because it is does not share parasympathetic influence (Braithwaite et al., 2013; Critchley, 2002). EDA has been linked to implicit emotional and attentional processing (e.g., responses to threat, anticipation, salience, novelty) (Braithwaite et al., 2013). Low resting EDA has been associated with psychopathy, sociopathy, and conduct problems, while increased EDA task reactivity has been associated with aggression, although findings vary by age and stimulus valence (Lorber, 2004).

ANS relations with psychological symptoms.

While much research has focused on RSA at rest, a meta-analysis showed small, significant effect sizes for the relation between RSA withdrawal during challenge tasks and fewer internalizing, externalizing, and cognitive/academic problems in children (Graziano & Derefinko, 2013). This suggests that regulatory functions of the PNS are compromised in youth with externalizing and internalizing problems. Further, this relation was moderated by the type of sample, where the strength of the relation was significantly more negative (i.e., increased RSA withdrawal and fewer externalizing problems) in community samples as compared to clinical/at-risk samples (Graziano & Derefinko, 2013). Of note, children in clinical samples often show excessive RSA withdrawal (Beauchaine et al., 2007), highlighting the need to assess physiological factors in conjunction with level of risk and/or psychopathology.

Consequences of Trauma and ACEs

The contexts in which children develop are essential for understanding physiological contributions to risk and resilience. Children exposed to trauma, including physical, emotional, and sexual abuse, and physical and emotional neglect, have a significantly heightened risk for psychopathology as they progress into adolescence and adulthood (Edwards et al., 2003; Font & Berger, 2015; McLaughlin et al., 2012; Molnar et al., 2001). ACEs (e.g., physical, sexual, or emotional abuse; physical or emotional neglect; natural disaster; parental death; caregiver psychopathology, substance abuse, and criminal behavior), show a graded relation between the number of exposures and later health risk behaviors (e.g., substance use, obesity) and disease (e.g., heart disease, cancer) (Felitti et al., 1998). The prevalence of PTSD in youth exposed to trauma is approximately 16%, with girls exposed to interpersonal trauma demonstrating the highest rates (33%) and boys exposed to non-interpersonal trauma with the lowest rates (8%) (Alisic et al., 2014). PTSD is a significant burden on individuals and families and corresponds to staggering economic and societal costs (Kessler, 2000; Magruder et al., 2017). The high prevalence of ACEs coupled with evidence that children who experience trauma are at risk for psychopathology underscores an important public health problem and the need to clarify potential mechanisms linking trauma to psychopathology (US Department of Health & Human Services, 2020). Identifying risk and protective factors for youth can guide targeted intervention.

A recent meta-analysis showed that children exposed to maltreatment (i.e., abuse or neglect) behaviorally display and report experiencing high levels of negative emotions (i.e., sadness, hostility, fear), and behaviorally display low levels of positive emotions compared to controls (Lavi et al., 2019). These findings align with attachment, social information processing, and learning theories predicting that ACEs result in hyporeactivity to positive emotions, hyperreactivity to negative emotional stimuli, and increased expression of negative affect. The aggregate effect size for physiological reactivity was not significant, suggesting that exposure to maltreatment alone may not correspond to physiological dysregulation (Lavi et al., 2019). In children exposed to trauma, questions remain about how physiological reactivity is related to the development of psychopathology.

Trauma and ANS physiological dysregulation.

Despite significant risks, not all individuals who have been exposed to ACEs and/or traumatic events develop psychopathology (Baldwin et al., 2021). Similarly, findings of ANS function in trauma-exposed youth are mixed. ACEs and trauma are frequently linked to hyperreactivity of the ANS, though a growing body of work shows that hyporeactivity or blunted responses also occur (see Obradović, 2012). This may negate aggregate findings of dysregulation in youth. While ACEs have been linked to ANS dysregulation (Busso et al., 2017; Cook et al., 2012; Gunnar & Quevedo, 2007; Koopman et al., 2004; Leitzke et al., 2015; McLaughlin, Sheridan, et al., 2014), certain physiological profiles have been identified as protective (El-Sheikh & Whitson, 2006; Jenness et al., 2019; McLaughlin, Alves, et al., 2014; Porges, 2007), and others show marginal differences between youth with and without trauma (Shenk et al., 2014). A recent review found that the majority of studies report blunted ANS responses at rest and in response to stress tasks, and blunted SNS responses during stress tasks in maltreated youth, but highlighted the diversity of methods and samples, making consolidation of findings difficult (Young-Southward et al., 2020).

Physiological Reactivity in PTSD

In addition to intrusion symptoms (e.g., reexperiencing the traumatic event), avoidance of trauma-related stimuli or emotional numbing, negative mood or negative trauma-associated cognitions, dissociation, and depersonalization, alterations in arousal and reactivity are a core PTSD symptom (American Psychiatric Association, 2013). Currently, the development of PTSD in youth is assessed in two ways: children exposed to the cumulative effects of ACEs, often encompassing chronic, overlapping, and interpersonal stressors; and children who experienced a specific event resulting in threatened death or serious injury (e.g., motor vehicle accident). These bodies of work are often siloed, yet can provide important insight into the development of PTSD if dysregulation of a stress response system (e.g., ANS) underlies the development and course of PTSD regardless of trauma type. Identifying risk and protective factors for PTSD development is essential for mental health professionals to determine who may require intervention, and which factors can be targeted by clinicians in treatment.

The ANS is important in understanding how trauma manifests biologically and psychologically. While PTSD is by definition a disorder triggered by trauma, it is heterogeneous and involves complex interactions among biological factors. In conjunction with evidence of hyper- and hyporeactivity of the ANS after ACEs and trauma, two PTSD subtypes have been suggested: a dissociative (hyporeactive) subtype characterized by extreme inhibition of emotion, and a re-experiencing/hyperaroused subtype characterized by “under-modulation” of emotion (Lanius et al., 2010; Wolf et al., 2012). Exploring PTSS on a continuum versus solely comparing groups with and without PTSD may also yield important nuances and heterogeneity in autonomic responses. For example, young adults with sub-clinical symptom levels demonstrated elevated HR and skin conductance responses during a startle task, while those with significant symptoms showed blunted reactivity (D’Andrea et al., 2013). As reviewed above, SNS- and PNS-mediated responses may contribute differently to these distinct PTSD presentations.

In adults, two meta-analyses have shown that elevated HR was significantly positively associated with PTSD, regardless of task type (i.e., rest, startle, and trauma cue paradigms), and demonstrated large effects (Nagpal et al., 2013; Pole, 2007), while the relationship was large and negative for PNS measures (Nagpal et al., 2013). A third meta-analysis demonstrated that the relation between HR and subsequent PTSS was small and positive across all ages, though in moderator analyses, the association remained positive for youth samples, yet was negative for older samples (Morris et al., 2016). In a fourth meta-analysis, there was a significant small, negative association between resting RSA and PTSD, though moderator analyses revealed that this relation was only significant for adult samples and not child/adolescent samples (Campbell et al., 2019). These meta-analyses highlight that the relation between specific ANS measures and PTSS in adults and in youth show disparate effects. Generalization of physiological features found in adults to children may not be appropriate given developmental experiences, maturation, and social and environmental factors.

Summary and Goals of Current Review

To summarize, the ANS is implicated in emotions, stress responses, and psychopathology. ACEs and trauma are often associated with physiological dysregulation and psychopathology. Increased understanding is needed of the relation between trauma and ANS function, and contributions of the ANS in PTSD in children and adolescents. Physiological measures have been assessed in relation to PTSS in adults (Nagpal et al., 2013; Pole, 2007), and singular specific indices (e.g., RSA) have been assessed in children in relation to PTSS (Campbell et al., 2019) and adaptive functioning or other psychopathology (Beauchaine, 2001; Graziano & Derefinko, 2013; Lorber, 2004; Shahrestani et al., 2014).

The present meta-analysis provides a quantitative review of the relations between various ANS measures and PTSS in youth to replicate and extend the current literature. The present study includes both combined ANS (i.e., having both SNS and PNS influence) and distinct influences of ANS branches (i.e., primarily SNS and primarily PNS measures), well-validated PTSS measures, and relevant moderators of the relation between psychophysiology and PTSD in child and adolescent samples. The relation between the ANS and outcomes was focused on PTSS as opposed to focusing solely on diagnoses of PTSD or internalizing disorders broadly in an effort to increase understanding of trauma- and stressor-related disorders specifically. The primary aims were to (a) determine the magnitude and direction of associations between ANS function and PTSS; (b) determine if associations vary for the SNS and PNS; (c) assess if age and sex moderate associations between ANS activity and symptoms; and (d) qualitatively review study methodologies and sample characteristics (e.g., task type, stressor characteristics) as potential sources of variability to be explored in future research. In alignment with meta-analyses of adults and the general consensus that PTSD is characterized by tonic autonomic hyperarousal, it is hypothesized that increased ANS and SNS activity during rest and stress will correspond to increased PTSS, while decreased PNS activity during rest and stress will correspond to increased PTSS. While many effects reporting on general ANS and SNS measures in adults have been large, research in youth has been mixed. Therefore, the magnitude of these associations in youth are expected to be small-to-medium effects.

Method

Literature Search

PsycINFO and PubMed were searched for empirical studies reported through July 1^st 2021 to identify articles examining ANS function and PTSS in children and adolescents using validated assessments. There was no lower limit for publication date to include all viable studies assessing measures of interest. The systematic literature search used the specific search terms (PTSD OR posttraumatic or post-traumatic) AND (autonomic nervous system OR sympathetic OR parasympathetic OR heart OR heart rate variability OR vagal OR electrodermal OR skin) AND (child OR adolescent OR pediatric) across all fields (i.e., title, abstract, keywords). Database searches were supplemented with backward searches reviewing the reference sections of published meta-analyses including trauma, PTSS, and ANS function.

Inclusion and Exclusion Criteria

Empirical studies were included if (a) the sample was comprised of children or adolescents (defined as having a mean age < 20 years); (b) utilized a valid and reliable measure of PTSS or PTSD; (c) assessed at least one measure of ANS, PNS, or SNS physiology; and (d) reported the quantitative relation between symptoms of PTSD or PTSS and ANS physiology. Articles were excluded if they (a) included a sample with a mean participant age above 20 years; (b) were not published in English; (c) reported statistics incompatible with meta-analytic aims; (d) involved experimental manipulation of PTSS or physiological measures (e.g., intervention studies), although if pre-intervention correlates of PTSS were reported, these were included in analyses; (e) did not include measures of interest or reported incompatible measurements (e.g., physiological activity measured during conditioning tasks or nocturnally; only reported salivary alpha amylase); or (f) reported a measure of only one dimension of PTSS (e.g., dissociation); or (g) the full-text article was unavailable (e.g., embargo, poster).

Study Selection

PsychINFO and PubMed searches yielded 1,298 studies and backward searches yielded an additional 92 studies. Duplicate records (N = 200) were removed, leaving 1,190 records to screen (Figure 1). Record titles and abstracts were screened for studies that assessed PTSD/PTSS and a measure of ANS physiology. If there was any doubt about eligibility, the record was retained for further review. During screening, 947 records were excluded, leaving 243 studies to assess for eligibility. If a dissertation and published version of the same sample were identified, the peer-reviewed published version was retained. Authors were contacted for additional data if articles did not include sufficient information to calculate effect sizes (k = 7); three authors responded and provided sufficient information, four authors did not provide data allowing for inclusion. Two independent raters completed all full-text reviews. Rater agreement across all articles reviewed for inclusion was 98%. Discrepancies were discussed until consensus was attained. Based on inclusion criteria, 38 records were included after full-text review. From these 38 studies, there was a total of 3,488 participants and 194 effect sizes were extracted.

Figure 1.

PRISMA Flow Diagram

Data Coding Procedure

The following information was extracted where available: (a) sample size; (b) summary statistics for the calculation of effect sizes (e.g., correlations); (c) sample demographic variables (i.e., mean sample age, percentage female, percentage White); (d) study type (i.e., cross-sectional, longitudinal); (e) task type during physiological data collection (i.e., rest or neutral task, stress task); (f) PTSD/PTSS symptom measure; (g) PTSD/PTSS measure informant (e.g., child, parent) and format (e.g., questionnaire or interview); (h) sample trauma characteristics (e.g., physical abuse, motor vehicle accident, mixed). One study reported both cross-sectional and longitudinal effects within the same publication, and was therefore included in both analyses of cross-sectional and longitudinal studies with cross-sectional and longitudinal correlations entered separately (Mikolajewski & Scheeringa, 2018). One study reported cross-sectional and longitudinal effects in two separate publications with the same sample (Shenk et al., 2012, 2014). One research group reported cross-sectional and longitudinal effects in two separate publications, with nearly identical samples (Zatzick, Grossman, et al., 2006; Zatzick, Russo, et al., 2006). Both cross-sectional and longitudinal effects were entered separately in these instances.

ANS measures were categorized as reflecting overall ANS activity (i.e., reflective of both SNS and PNS activity), primarily PNS activity, and primarily SNS activity based on prior research (see Appendix A for categorizations). Of note, two studies reported HRV measures (Haag et al., 2019; MacArthur, 2011). In the present study, high frequency HRV was categorized as a PNS measure, while low frequency HRV and high frequency to low frequency ratio were categorized as ANS measures (i.e., with both SNS and PNS influence). These categorizations were made in accordance with current research findings, although it is important to note that research on these categorizations is ongoing and the primary index of various HRV frequencies continues to be debated (Berntson, et al., 1997; Shaffer & Ginsberg, 2017; Heathers, 2014). Task types were categorized as either rest/neutral or stress tasks. Rest/neutral tasks were categorized by measures taken while participants were instructed to rest (i.e., with no stimuli presented) or if presented with neutral or non-emotional stimuli (e.g., neutral video). Stress tasks were categorized by measures taken while participants viewed emotional or stressful stimuli (e.g., violent video, personal trauma recall, conflict discussion task) or if they were taken during a period immediately following a stressful or traumatic experience (i.e., administered immediately after injury or upon emergency department arrival). A primary rater extracted all study data. A secondary rater double-extracted data for 52% of included studies (k = 20). Inter-rater agreement across double-extracted data was 98%, and all disagreements were discussed and corrected during consensus meetings.

Computation of Effect Sizes

As the associations between continuous variables were examined (i.e., continuous measures of PTSS and of ANS, SNS, and PNS function), correlation coefficients (r) were the effect sizes utilized to quantify the relation between PTSS and ANS function for the present review. Non-significant findings reported without exact values specified (i.e., “all other correlations were nonsignificant”) were conservatively estimated and coded as a correlation of zero (Lipsey & Wilson, 2001). If correlations were not reported, they were estimated based on univariate statistics (e.g., means and standard deviations, F-test statistics, Chi-square statistics, and p-values). For studies that only reported standardized βs, we applied Peterson and Brown’s (Peterson & Brown, 2005) formula (r = β + 0.05 λ [where λ = 1 for non-negative βs, and λ = 0 for negative βs]) to impute the corresponding r values. A threshold of a minimum of 3 studies (k = 3) was used to estimate effect sizes for each category of analyses.

Effect sizes from the same study (i.e., a single study reporting correlations with the same correlate across different measures of the other domain) are not independent (Borenstein et al., 2009). Therefore, the mean effect size (r) for each study was used as the level of analysis, such that each study included would contribute only one summary effect size to the main analysis. Two of the identified articles included the same sample or a smaller subset of the same sample published in multiple papers, and these articles reported separate cross-sectional and longitudinal effects. Therefore, adjustments for overlapping samples were not necessary.

Supplemental analyses.

The present meta-analysis included studies that assessed PTSS on a continuum, as well as studies comparing participants with PTSD diagnosis versus controls. Combining these studies reflects current thinking in psychopathology. It allows for assessment of the ANS-PTSS relation along a continuum, and to include participants who may fall short of diagnostic criteria, but could be experiencing significant distress or impairment nonetheless (Kotov, et al., 2017). Despite this, it is plausible that ANS-PTSS relations vary as a function of PTSD diagnosis, where symptoms reach a diagnostic classification cutoff. Therefore, we report additional results for studies reporting ANS differences between participants with a PTSD diagnosis versus controls (k = 11) for each set of analyses.

All analyses were conducted with the Comprehensive Meta-Analysis Program Version 3 (Borenstein et al., 2014). Because included studies contained noticeably different features and samples, we used random-effect models rather than fixed-effect models, as fixed-effects models assume that all included studies are functionally identical and share a single canonical effect size (Borenstein et al., 2010; Hedges & Vevea, 1998). Additionally, the random-effects model allows unconditional inferences (i.e., a generalizable conclusion to situations beyond the sampled studies) of the results (Field, 2001). Cohen’s (1988) guidelines were used to interpret the magnitude of the effect size for correlations (i.e., r = .10 represents a small effect, r = .30 a medium effect, and r ≥ 50 a large effect). Weighted mean effect sizes (r), 95% confidence intervals (CIs), and estimated heterogeneity statistics (Q and I²) were calculated using procedures outlined by Hedges and Olkin (1985). The 95% CIs on effect sizes represent the range in which the mean effect size will be in 95% of cases. All p values were two-tailed.

Publication bias

Publication bias can lead to inflated or incorrect estimates of effects. Bias was first addressed by including unpublished dissertations. Second, funnel plots of effect sizes plotted against the standard error were visually examined. Third, potential publication bias was empirically evaluated by calculating Egger’s tests to detect funnel plot asymmetry (Egger et al., 1997). Fourth, Trim-and-Fill analyses were conducted to determine how many studies would need to be included to achieve funnel plot symmetry (Duval & Tweedie, 2000; Shi & Lin, 2019). Finally, Begg and Mazumdar (Begg & Mazumdar, 1994) rank correlation tests were used to assess the correlation between study size and effect size.

Moderator Analyses

To analyze effects of continuous moderators, meta-regression analyses were conducted to determine if moderators accounted for unexplained variability across samples. Moderator analyses were limited to effects with significant heterogeneity as reflected in the Q and I statistics in order to determine if variability could be explained by moderators in those cases. Age and sex were chosen as primary moderators of effects, as there may be differences in the relation between ANS measures and PTSS across development (Campbell et al., 2019; Morris et al., 2016) and there are well-established differences in the prevalence of PTSD by sex in youth (Alisic et al., 2014). Age was assessed as a continuous variable (i.e., mean sample age), as was sex (i.e., percentage of the sample identifying as female).

Results

Study Characteristics

A total of 38 studies were included in quantitative analyses (N = 3,488). Included studies had a mean sample age of 13.11 years, ranging from 4.98 to 19.55 years. Twelve samples assessed children (ages one to 10 years), 15 assessed youth in early adolescence (ages 11 to 15), and 11 in late adolescence (ages 16 to 20). The mean proportion of female participants across studies was 51.23%. The mean proportion of White participants was 49.52%. The majority of studies (k = 24) were cross-sectional and 13 were longitudinal; one study reported both cross-sectional and longitudinal effects (Mikolajewski & Scheeringa, 2018). Nineteen studies reported only ANS measures, eight reported only PNS measures, and one reported only SNS measures. Nine studies reported data on multiple physiological systems: five studies reported both ANS and SNS measures, three reported both PNS and SNS measures, one reported ANS and PNS measures, and one reported on all three (i.e., ANS, PNS, SNS). See Appendix B for complete study descriptions and sample characteristics.

Association of Combined ANS Measures with PTSS

Baseline rest/neutral tasks.

Across all study types (i.e., cross-sectional and longitudinal) (k = 12), the relation between ANS (i.e., combined PNS and SNS influence) measures during rest tasks and PTSS was non-significant, r = .13, p = .19 (Table 1). Heterogeneity was significant, as shown by the Q statistic (Q = 173.96, p < .001), and I² = 93.68. When assessed only in samples comparing those with and without PTSD diagnosis (k = 7), the correlation remained small and positive, r = .21, p = .08, and was non-significant. Examining only cross-sectional studies (k = 10), the relation between ANS physiology and PTSS was non-significant (p = .31), and there was significant heterogeneity, Q = 160.90, I² = 94.41). This finding was similarly non-significant when assessed only in samples comparing those with and without PTSD diagnosis. Only two studies reported longitudinal effects for the relation between ANS physiology and PTSS utilizing a rest task, therefore meta-analyses were unable to be completed.

Table 1.

Effects for Correlates of ANS Measures During Baseline Rest/Neutral Tasks and PTSS

	ANS					PNS						SNS
	k	N	r [95% CI]	Q	I 2		k	N	r [95% CI]	Q	I 2	k	N	r [95% CI]	Q	I 2
All Studies	12	1,132	.13 [−.06, .31]	173.96	93.68		12	1,124	−.01 [−.07, .05]	13.13	16.19	7	411	−.09 [−.17, −.007]	5.26	0
CS Samples	10	980	.11 [−.11, .32]	160.90	94.41		8	787	−.01 [−.09, .07]	9.11	23.12	7	411	−.09 [−.17, −.007]	5.26	0
L Samples	2	152	--	--	--		5	363	−.02 [−.13, .10]	4.27	6.34	0	--	--	--	--

Note. CS = cross-sectional; L = longitudinal; ANS = autonomic nervous system; PNS = parasympathetic nervous system; SNS = sympathetic nervous system; k = number of independent samples; n = number of youth in sample; r = aggregated correlation coefficient; 95% CI = 95% confidence interval of correlation coefficient; Q = Q statistic for total within heterogeneity; I² = statistic for heterogeneity. Bolded text indicates significant effects; significance for the Q statistic involves a statistical test and p-value, while I² values above 20% to 40% are considered indicative of important heterogeneity.

-- Signifies insufficient studies to conduct analyses for these cells.

Stress tasks.

Across all study types (k = 22), there was a significant positive relation between ANS measures during stress tasks and PTSS, r = .07, p = .04, 95% CI = .003 to .14, a small effect (Table 2). There was significant heterogeneity in the model, as shown by the Q statistic (Q = 54.73, p < .001) and I² = 61.63. When assessed only in samples comparing those with and without PTSD diagnosis (k = 5), the correlation remained small and postive, r = .11, p = .20, and was non-significant. Assessing only cross-sectional studies (k = 13), this effect was no longer significant, p = .91, and there was significant heterogeneity (Q = 29.14, p = .004, I² = 58.83). This finding was similarly non-significant when assessed only in samples comparing those with and without PTSD diagnosis. In longitudinal studies (k = 9), there was a significant positive association between ANS measures during stress tasks and PTSS, r = .15, p < .001, 95% CI = .07 to .22, a small effect, and significant heterogeneity in the model, Q = 15.73, p = .05, I² = 49.14. Only one study reported longitudinal effects comparing those with and without PTSD diagnosis for this effect, therefore, this was unable to be assessed meta-analytically.

Table 2.

Effects for Correlates of ANS Measures During Stress Tasks and PTSS

	ANS					PNS					SNS
	k	N	r [95% CI]	Q	I 2	k	N	r [95% CI]	Q	I 2	k	N	r [95% CI]	Q	I 2
All Studies Adjusted Value	22	1,512	.07 [.003, .14] .02 [−.05, .10]	54.73 94.29	61.63 --	9	623	−.07 [−.19, .05]	28.16	71.59	9	466	−.08 [−.1, .16]	77.81	89.72
CS Samples	13	687	.006 [−.09, .10]	29.14	58.83	6	393	−.04 [−.16, .08]	10.68†	53.20	8	422	−.07 [−.33, .19]	77.69	90.99
L Samples	9	825	.15 [.07, .22]	15.73	49.14	4	256	−.09 [−.29, .12]	13.93	78.46	1	44	--	--	--

Note. CS = cross-sectional; L = longitudinal; ANS = autonomic nervous system; PNS = parasympathetic nervous system; SNS = sympathetic nervous system; k = number of independent samples; n = number of youth in sample; r = aggregated correlation coefficient; 95% CI = 95% confidence interval of correlation coefficient; Q = Q statistic for total within heterogeneity; I² = statistic for heterogeneity. Bolded text indicates significant effects; significance for the Q statistic involves a statistical test and p-value, while I² values above 20% or 40% are considered indicative of important heterogeneity. Italicized values reflect values adjusted for effect sizes in which trim-and-fill analyses indicated potential publication bias.

-- Signifies insufficient studies to conduct analyses for these cells.

^†p = .06

Association of PNS Measures with PTSS

Baseline rest/neutral tasks.

Across all study types (k = 12), the relation between PNS activity during rest tasks and PTSS was non-significant, p = .74, and heterogeneity statistics were non-significant, Q = 13.13, p = .29, I² = 16.19 (Table 1). In cross-sectional studies (k = 8), the relation was non-significant, p = .81, with no evidence of heterogeneity, Q = 9.11, p = .25, I² = 23.12. In longitudinal studies (k = 5), the relation was also non-significant, p = .74, as were heterogeneity statistics, Q = 4.27, p = .37, I² = 6.34. There were not enough studies to calculate these effects only in samples reporting differences between those with and without PTSD.

Stress tasks.

Across all studies (k = 9), the relation between PNS activity measured during stress tasks was not significantly related to PTSS, p = .24, though there was significant heterogeneity, Q = 28.16, p < .001, I² = 71.59 (Table 2). For cross-sectional studies (k = 6), the relation between PNS activity during stress tasks and PTSS was non-significant, p = .51. The Q statistic approached significance, Q = 10.68, p = .06, while the I² statistic indicated potential heterogeneity (I² = 53.20). In longitudinal samples (k = 4), the correlation was also non-significant, p = .41, and indicated significant heterogeneity, Q = 13.93, p = .003, I² = 78.46. There were not enough studies to calculate these effects when contraining analyses to samples reporting differences between those with and without PTSD.

Association of SNS Measures with PTSS

Baseline rest/neutral tasks.

No studies assessing the effect of SNS activity during rest tasks and PTSS were longitudinal, therefore, only cross-sectional results are presented (k = 7). There was a significant negative relation between SNS activity during rest tasks and PTSS, r = −.09, p = .03, 95% CI = −.17 to −.007, a small correlation. There was no evidence of heterogeneity, Q = 5.26, p = .51, I² = 0.00 (Table 1). When constrained to samples with and without PTSD diagnosis (k = 3), the correlation was similarly small and negative, and was non-significant, r = −.11, p = .14, 95% CI = −.24 to .03.

Stress tasks.

Across all study types (k = 9), the relation between SNS activity during stress tasks and PTSS was non-significant, p = .52 (Table 2). There was significant heterogeneity in the model, Q = 77.81, p < .001, I² = 89.72. When constrained to samples assessing differences between those with and without PTSD diagnosis (k = 3), the correlation was medium, negative, and was non-significant, r = −.32, p = .39, 95% CI = −.79 to .40. In cross-sectional samples (k = 8), the relation between SNS activity and PTSS was also non-significant, p = .59, and there was significant heterogeneity, Q = 77.69, p < .001, I² = 90.99. One study reported longitudinal results, therefore effects could not be calculated. There were not enough studies to assess SNS measures during stress in cross-sectional and longitudinal samples separately when constrained to studies of those with and without PTSD diagnosis.

Moderator Analyses

Age.

Across all studies, mean sample age did not moderate the relation between PTSS and ANS measures during rest tasks, coefficient = .002, 95% CI [−.05, .05], p =.93 nor during stress tasks, coefficient = −.02, 95% CI [−.04, .002], p = .08. There was no effect of age on SNS measures during stress tasks and PTSS across all studies, coefficient = −.04, 95% CI [−.14, .06], p = .43. In analyses of cross-sectional and longitudinal studies separately, age similarly did not moderate the association between PTSS and ANS measures in rest or stress tasks, nor did age moderate the association between PTSS and SNS measures, p > .05. In contrast, the moderation effect for mean sample age on the association between PTSS and PNS function during stress tasks (k = 8) approached significance, coefficient = .04, 95% CI [−.0001, .07], p = .05, R² = .62, indicating that 62% of the variance in the relation was attributed to sample age. The meta-regression demonstrated that the correlation between PTSS and PNS activity was stronger and negative at younger ages, while it approached zero when assessed in older samples (see Figure 2). The moderating effect of mean sample age on the relation between PNS measures during stress tasks and PTSS was non-significant when assessed in cross-sectional samples and longitudinal samples, p > .05.

Figure 2.

Regression of Fisher’s Z on Sample Mean Age for PNS measures during stress tasks across studies.

Sex.

Across all studies, participant sex (i.e., percent of the sample identifying as female) did not moderate the association between PTSS and ANS measures during rest tasks, coefficient = .002, 95% CI[−.01, .01], p = .79, nor during stress tasks, coefficient = −.002, 95% CI [−.005, .0005], p = .12. Similarly, there was no effect of sex on the relation between PTSS and PNS measures during stress tasks across all studies, coefficient = .004, 95% CI [−.0003, .008], p = .07, nor SNS measures, coefficient = .003, 95% CI [−.01, .02], p = .76. In separate analyses of cross-sectional and longitudinal studies, sex similarly did not moderate the association between PTSS and ANS, PNS, or SNS measures, p > .05.

Publication Bias

Rest/neutral tasks.

There was a significant effect size for SNS activity during rest tasks and PTSS cross-sectionally. Visual assessment of the funnel plot was not suggestive of bias (Figure 3). The Egger’s test and Begg and Muzumdar rank correlations were non-significant, p > .05. Trim-and-fill analyses showed that no studies were trimmed. Taken together, analyses did not indicate susceptibility to publication bias.

Figure 3.

Funnel plots of the relationship between the standard error and Fisher’s Z in studies of physiological measures and PTSS. (a) ANS measures during stress tasks across all studies. (b) ANS measures during stress tasks in longitudinal samples. (c) SNS measures during rest tasks in cross-sectional samples.

Stress tasks.

There was a significant positive effect for the relation between ANS measures during stress tasks and PTSS across all studies. Visual assessment of the funnel plot was not suggestive of publication bias (Figure 3). This effect produced a non-significant Egger’s test (regression intercept = −.57, 95% CI [−3.07, 1.93]), p = .64, two-tailed. Duval and Tweedie’s trim and fill analyses revealed that zero studies to the right of the mean needed to be trimmed, though four studies trimmed and filled to the left of the mean. When accounting for potential publication bias, the result was no longer significant, adjusted value = .02 (95% CI [−.05, .10]). Begg and Mazumdar rank correlations were non-significant (i.e., not indicative of publication bias), p > .05. Of note, there was no indication of publication bias for this same significant effect in solely longitudinal analyses. Visual assessment of the funnel plot in longitudinal samples was not suggestive of publication bias (Figure 3). The Egger’s test and Begg and Mazumbdar rank correlations were non-significant, p > .05. Trim-and-fill analyses showed no studies trimmed on either side of the mean.

Qualitative Review

It is possible that effects could be moderated by additional factors that could not be addressed in the current meta-analysis but could be addressed in future research. These factors are now considered in a qualitative review.

Sample and stressor characteristics.

Included studies varied greatly in terms of sample characteristics and stressors that youth experienced (See Appendix C). However, there were insufficient studies to quantitatively examine this in ANS, PNS, and SNS analyses. Four studies included in the present meta-analysis recruited community samples without any trauma history. Seven studies included participants who had experienced a physical injury (i.e., motor vehicle accidents, falls, burns, animal attacks, and sports injuries). Three studies included samples of youth exposed to specific traumatic events, including living in a metropolitan area at the time of a terrorist attack, exposure to a mass campus shooting, and youth of parents who experienced a bombing. One study included youth predominantly exposed to either natural disasters, transportation accidents, or near drowning. Another study included a group of war-exposed youth residing in area with frequent and unpredictable rocket attacks.

Further, 13 studies reported varied types of childhood maltreatment. These samples included children who had experienced physical abuse, physical neglect, sexual abuse, interpersonal or domestic violence, community violence, loss of a parent, or a combination of multiple interpersonal traumas. Of note, an additional nine studies included samples of youth who had exposure to both physical injury or disaster and interpersonal trauma. It may be that the relation between the ANS and PTSS demonstrates a physiological profile characterized by hyperreactivity in youth with intense, single-instances of trauma, while those exposed to multiple, chronic, and/or interpersonal trauma demonstrate a hyporeactive profile (McEwen, 1998; National Scientific Council on the Developing Child, 2020).

Study methodology.

Studies included in the present meta-analysis varied in their study design and measures used (See Appendix B). The majority of studies (k = 19) assessed baseline rest/neutral measures of the ANS with simple instructions for participants to rest (i.e., no stimuli shown), while some studies (k = 8) recorded physiological activity while participants watched or listened to a neutral story or video, or completed a simple writing task. Analyses of ANS baseline rest/neutral measures demonstrated significant heterogeneity, therefore, it is possible that differences in rest/neutral measurements contributed to variability for this effect.

All of analyses of stress tasks indicated significant heterogeneity. Differences in the type of stress task completed across studies is one plausible explanation for this variability. Stress recovery and stressful laboratory tasks were combined in quantitative analyses, though may differ in important ways (e.g., timing from trauma, type of trauma). Nine studies were classified as stress recovery in the present meta-analysis. This included measurement of physiological activity soon after a traumatic event (e.g., taken at the scene of an accident, during emergency medical services transport, at emergency department arrival/triage or admission, 24 hours post-admission). While these studies did not present stressful stimuli to participants, they were coded as stress measurements because they were taken soon after a traumatic event. Within stress laboratory tasks (k = 20), there was substantial variability in task type, study procedures, and methods. While quantitative analyses of task type were beyond the scope of this review, this may have contributed to variability in findings. Specifically, physiological measures taken soon after trauma (i.e., stress recovery) may demonstrate a more consistent pattern. Stress recovery measures may reveal a profile characterized by hyperreactivity, as these were often non-interpersonal, intense, individual instances of stress. Stress laboratory tasks would likely demonstrate more variability in physiological responses, as tasks varied from non-interpersonal (e.g., startle tasks), interpersonal but non-dyadic (e.g., Trier Social Stress Task), interpersonal and dyadic (e.g., parent-child conflict discussion), or personalized (e.g., interview about one’s own traumatic experience). These different stimuli include both within- and between-subjects factors. For example, trauma-related or personalized stimuli may engage memory systems in a distinct way, which may in turn affect patterns of stress responses. Additional research is needed to delineate any potential differerences in physiological profiles across stress tasks.

Another potential source of variability in findings includes the measurement of PTSS. Approximately half of included studies (k = 18) used questionnaire measures, approximately half (k = 17) utilized interviews, and three studies used multiple measures to assess symptoms in youth (See Appendix C). It is encouraging that the majority of included studies used rigorous symptom assessments (e.g., diagnostic interviews) or multiple reports. In addition, PTSS informant could contribute to variability across findings. While the majority of studies (k = 26) relied solely on youth self-reported symptoms, some studies utilized multiple informants to assess symptoms (k = 8) or solely caregiver report (k = 4). Symptoms reported by caregivers may not capture the full range of internal experiences of PTSD in youth.

Discussion

The current meta-analysis and review of 38 studies quantitatively reviewed the relations between multiple measures of ANS function (e.g., combined PNS and SNS influence, primarily PNS, and primarily SNS) and PTSS in youth in cross-sectional and longitudinal samples. ANS measures have been assessed in relation to PTSS in adults (Nagpal et al., 2013; Pole, 2007), and discrete ANS indices (e.g., resting RSA) have been assessed in children in relation to PTSS (Campbell et al., 2019). The findings of the current meta-analysis extend the literature on autonomic function and the onset and development of trauma- and stressor-related disorders. Included studies’ samples were representative across sex, race, and age. Results showed a small, positive relation between ANS measures and PTSS during stress tasks across all studies and longitudinally. SNS activity during rest tasks demonstrated a small negative correlation with PTSS in cross-sectional samples. The effect for age moderating the relation between PNS activity measured during stress tasks and PTSS approached significance, such that the correlation was stronger in younger samples. The findings suggest that the relations between overall measures of ANS function and SNS-specific measures with PTSS, but not PNS activity, are most supported by the current state of research.

Autonomic Dysregulation: Hyperreactivity and Hyporeactivity

According to the allostatic load model, an adaptive physiological response to stress is initiated, sustained for a period of time, and subsequently decreases as systems recover back to baseline levels (McEwen, 1998). Brief and intermittent activation of stress systems followed by a return to baseline can build adaptation and resilience, but if continually activated due to intense or prolonged stress, systems will experience allostatic wear and tear that can undermine health (McEwen, 1998; National Scientific Council on the Developing Child, 2020). Physiological dysregulation may be reflected in to two different profiles. Some may show initial activation of physiological stress systems with delayed recovery (i.e., a prolonged response) (McEwen, 1998), which is consistent with evidence that those with PTSD demonstrate hyperreactivity (Nagpal et al., 2013; Pole, 2007). Second, it may be that when exposed to a stressor, physiological stress responses are inadequate, demonstrating a blunted pattern, consistent with reports that some individuals with PTSD demonstrate hyporesponsivity to stress. The present analyses provide initial evidence in support of both profiles in youth: increased non-specific ANS activity in response to stress and decreased SNS activity at rest were associated with increased PTSS.

ANS.

While these effects were small in magnitude, a consistent finding in quantitative analyses indicated that ANS physiology measured during stress was positively correlated with PTSS in youth when combined across all study types and when assessed in longitudinal samples. Results were non-significant for cross-sectional designs. There was some indication of publication bias in the trim-and-fill analyses when combined across all study types, though funnel plots were not skewed and all other quantitative bias indicators were non-significant. While trim-and-fill analyses are recommended for meta-analyses, it is suggested to interpret these with caution in studies with significant heterogeneity, as in the present findings (Shi & Lin, 2019). Further, publication bias was not indicated when constrained to a more rigorous study design (i.e., longitudinal). Some studies assessed this relation longitudinally measured ANS function initially and PTSS subsequently. Therefore, the relation between ANS activity during stress and PTSS may emerge or strengthen over time, supporting hypotheses that increased ANS activity may be a marker of later symptom development (Kirsch et al., 2011). Increased ANS activity in youth found in the present analyses aligns with previous research (Morris et al., 2016; Nagpal et al., 2013; Pole, 2007), and the generally accepted notion that PTSD corresponds to high physiological arousal. As general ANS measures reflect features of both PNS and SNS function, findings suggest that general dysfunction of either or both parasympathetic and sympathetic systems corresponds to increased PTSS. Of note, many studies used general ANS measures, increasing power to detect effects. These measures (e.g., HR, BP) may be most easily measured in youth, as they are quick and uncomplicated to obtain with little to no discomfort.

It cannot be determined if autonomic dysregulation precedes or follows the onset of PTSS from the present analyses. Some studies assessed ANS physiology prior to measurement of PTSS and demonstrated that heighted ANS reactivity to stress corresponded to later increased PTSS for youth who experienced physical injuries (Bryant et al., 2007; De Young et al., 2007). Results reported by Haag et al. (Haag et al., 2019) demonstrated that resting HR measured 1-month post-injury was not correlated with PTSS concurrently or longitudinally, while HR reactivity in response to trauma narratives was consistently related to PTSS cross-sectionally and longitudinally. Future studies should continue to assess these measures longitudinally to determine if heightened ANS activity corresponds to subsequent increased PTSS or vice versa. It is also important to consider the feasibility of assessing physiology in close proximity to traumatic experience(s) in youth exposed to multiple traumas or maltreatment, even if assessed longitudinally, compared to one-time physical injuries resulting in emergency service utilization.

SNS.

In rest tasks, SNS activity demonstrated a small negative association with PTSS, where lower resting SNS activity corresponded to higher symptoms, contrary to what is generally accepted about PTSD (i.e., higher arousal). Of note, measurements of the SNS at rest were not the main focus of included studies; they were often solely used as comparisons for stress tasks. While this effect was small in magnitude, the present results demonstrate that resting SNS measures deserve more careful consideration. While SNS activity during stress was not associated with PTSS in the present analyses, increased SNS activity at rest may align with theories that responses to extreme or prolonged stress after trauma may reflect more “defensive” responses, including blunted arousal or non-reactivity (McEwen, 1998; Porges, 2003b, 2007) and research showing that some youth exposed to trauma demonstrate blunted physiological responses (Busso et al., 2017; D’Andrea et al., 2013; Leitzke et al., 2015; McLaughlin, Sheridan, et al., 2014). For example, young children who have experienced high levels of threat (vs. deprivation) exposure demonstrated blunted SNS activity in a fear conditioning task (Machlin et al., 2019). In college students, individuals with significant symptoms of PTSD and early exposure to multiple types of trauma demonstrated decreased SNS activity to startling sounds (D’Andrea et al., 2013). Of note, the opposite pattern of results (i.e., increased SNS activity) was found for individuals with sub-clinical PTSS, indicating variability in physiological responses based on the severity of trauma exposure and symptom progression. This should continue to be assessed in relation to proposed PTSD subtypes (i.e., dissociative subtype characterized by extreme inhibition of emotion, and a re-experiencing/hyperaroused subtype characterized by “under-modulation” of emotion (Lanius et al., 2010; Wolf et al., 2012). Future work should assess the relation between ANS measures and PTSS symptom clusters (e.g., arousal, hypervigilance, dissociation/numbing) to inform potential PTSD subtypes.

PNS.

Contrary to hypotheses, there were no significant PNS findings. This aligns with previous research suggesting that PNS measures are only related to PTSS in adults (Campbell et al., 2019; Nagpal et al., 2013) and research showing that youth with and without trauma exposure and with and without PTSD do not differ on PNS measures across pleasant or traumatic recall tasks (Gray et al., 2018). In contrast, others have shown a relation between RSA and PTSS (Mikolajewski & Scheeringa, 2018) and internalizing psychopathology in youth (McLaughlin, Alves, et al., 2014; McLaughlin et al., 2015). Polyvagal theory posits that RSA withdrawal is an index of emotion regulation (Porges, 2007) and high resting RSA corresponds to quicker physiological adaptation to stress (Lane et al., 1992). Some posit that low resting RSA may be a clinical marker of stress sensitivity predicting psychopathology after stress exposure (McLaughlin, Sheridan, et al., 2014; Mikolajewski & Scheeringa, 2018). The present findings do not support a significant relation between PNS function and PTSS in youth. Other ANS measures may be more sensitive to PTSS. Similar to the present meta-analysis, Cohen et al. created composite indices of PNS and SNS measures, and their sympathetically-driven index corresponded to PTSS in youth and discriminated between participants with and without PTSD, while PNS reactivity alone was non-significant (Cohen et al., 2020). Additionally, the majority of included studies reported RSA in relation to symptoms. RSA plays a central role in regulating energy exchange via synchronization of respiratory and cardiovascular processes durings metabolic and behavioral changes (Grossman & Taylor, 2007). Importantly, age, posture, respiration rate, and physical activity can confound the relation between vagal tone and RSA (Grossman & Taylor, 2007). Therefore, it is important that future studies account for these confounds in their measurement and interpretation of RSA (see Grossman & Taylor, 2007, for discussion of ways to address some of these methodological confounds).

As PNS responses are arguably integrated into the social engagement system (Porges, 2003a, 2007), it may be that the type of task (e.g., social context) is particularly relevant for PNS measures. Therefore, there may be important moderators of the relation between PNS measures and PTSS, or the PNS may moderate the relation between a third variable and PTSS. For example, exposure harsh parenting during childhood was associated with PTSS in adulthood for females with higher resting RSA and higher interparental aggression exposure (Barry et al., 2015). In a parent-child conflict task, RSA reactivity was negatively related to PTSS in youth, suggesting that less PNS control was associated with increased symptoms (Cohen et al., 2020). They found no associations during rest measures. In a meta-analysis, Shahrestani et al. (Shahrestani et al., 2014) found that in healthy children, HRV did not vary during social engagement tasks (i.e., free play, dyadic teaching tasks) compared to rest. HRV was reduced during social disengagement tasks (e.g., separation from parents), while reunion resulted in a return back to baseline levels in controls. In contrast, children either at risk for or diagnosed with psychopathology did not show HRV changes across social contexts. This could reflect a lack of autonomic flexibility in social tasks for at-risk children. In analyses of nondyadic social stress, there was a large effect for HRV reductions in children regardless of psychopathology status (Shahrestani et al., 2014). Overall, research on the relation between PNS measures and PTSS is inconsistent. ANS responses during dyadic social interactions specifically could reflect an important biomarker of psychopathology or resilience in youth.

General Dysregulation of PNS and SNS systems.

Initially, it may seem counterintuitive that ANS measures reflecting both PNS and SNS influence during rest tasks demonstrated a positive relation with PTSS, while SNS measures demonstrated a negative relation with PTSS, and no significant results were found for PNS measures. It is noteworthy that only one study (Kirsch et al., 2015) included in the present meta-analysis reported all three physiological domains in the same sample. While SNS and PNS responses are coordinated to flexibly meet environmental demands, the relation between them is complicated. Assessment of one aggregate measure cannot lead to meaningful understanding of complex processes contributing to typical or atypical psychological development (Beauchaine, 2001). SNS activation does not necessitate PNS activation or inactivation, and many physiological and behavioral responses represent coactivation of both systems in target organs (Levenson, 2014). Assessment of multiple measures in the same sample are needed to provide important insight into which physiological processes go awry in in youth with PTSD, and inform potential PTSD subtypes dominated by different physiological profiles.

As reviewed by Obradović (2012), trauma is frequently linked to ANS hyperreactivity, though a growing body of work shows that hyporeactivity also occurs. The present findings suggest ANS hyperreactivity to stress and SNS hypoactivity at rest are associated with higher levels of PTSS. Therefore, given the state of current research, it may be most appropriate to infer only that trauma corresponds to dysregulation, and not to either hyper- or hyporesponsivity. If both hyper- or hyporeactive profiles occur after extreme stress exposure, research is needed to delineate how physiological reactivity corresponds to either maladaptive or adaptive functioning in both the short- and long-term, for which ANS branch, and depending on the child’s environment. If studies include some youth demonstrating hyperreactivity and others hyporeactivity, effects may not appear at the aggregate level and yield null results. Studies assessing profiles of symptoms in youth with PTSS utilizing latent profile or latent class analyses could determine if there are PTSS subtypes where some youth demonstrate patterns of hyperreactivity while others demonstrate hyporeactivity of different ANS measures. Further, the relation between various ANS measures and PTSS may be nonlinear or be moderated by individual-level and measurement-level variables (Shaffer & Ginsberg, 2017).

Moderators of Effects

Age/Developmental considerations.

Physiological features of PTSD in adults may not generalize to children given developmental differences in ANS function, and social and environmental factors. Further, frequent or intense stress responses increase the risk of physical and mental health problems, particularly during periods of rapid child development (Danese & McEwen, 2012; Gunnar & Quevedo, 2007). In the current analyses, the moderating effect of sample age on PNS indices measured during stress tasks approached significance, such that the relation between PNS activity and PTSS was negative in child and young adolescent samples, and effects were small to negligible in older adolescent samples. Research on PNS development has been mixed: RSA has been shown to increase with age (Gray et al., 2018; Hinnant et al., 2017), while others have shown stability in these measures in youth (El-Sheikh, 2005). It is possible that younger children do not have as well-developed PNS-mediated regulatory abilities, and therefore are more susceptible to PTSS compared to adolescents and adults. Additional research is needed in order to understand age differences considering biological development.

A previous meta-analysis revealed a significant small, negative association between resting RSA and PTSD, though this relation was only significant for adult samples and not child/adolescent samples (Campbell et al., 2019). In adults, when comparing groups with PTSD and controls, participants with PTSD demonstrated lower vagal tone (Nagpal et al., 2013; Thome et al., 2017). The relation between PNS measures and PTSS may be more pronounced in adults or clinical samples. Current findings and previous research suggest that the relation between PNS measures and PTSS may be nonlinear and complex across development (Campbell et al., 2019; Morris & Rao, 2013). Older youth may have more developed regulatory systems but also may have more exposure to traumatic experiences (Cohen et al., 2020). Age differences in the relation between ANS measures and PTSS requires further research utilizing larger samples encompassing a wide age range in order to disentangle the contributions of autonomic functioning to PTSS across the lifespan.

Sex differences.

The relation between ANS measures and PTSS was not moderated by sex. Women and girls are more likely to develop PTSD than their male counterparts (Alisic et al., 2014; Tolin & Foa, 2006). Differences in ANS activity across sex has been mixed (Haag et al., 2019; Jenness et al., 2019; Koenig et al., 2017; Ordaz & Luna, 2012). It may be that other (and potentially interrelated) factors (e.g., learned fear responses, fear extinction, cognitive appraisals, posttraumatic cognitions, perceived helplessness, hormonal differences) contribute to the increased risk of PTSD in women and girls (Christiansen & Berke, 2020; Christiansen & Hansen, 2015; Gamwell et al., 2015; Shansky, 2015; Zoladz & Diamond, 2013). Additionally, sex may be confounded with trauma type. For example, females are more likely than males to experience sexual assault and child sexual abuse, while males are more likely to experience accidents, nonsexual assaults, witnessing death or injury, disaster or fire, and combat or war (Alisic et al., 2014; Tolin & Foa, 2006), and studies of sexual abuse are often conducted with predominantly female samples (Walker, Carey, Mohr, Stein, & Seedat, 2004). The present study included youth ages one to 20. It is plausible that sex differences may be more pronounced in older samples, similar to other internalizing disorders (Dalsgaard, et al., 2020). To assess the potential sex differences in the relation between ANS measures and PTSS, future studies should report effects separately for males and females.

Clinical Implications

Biological systems, including the ANS, are integrally involved in the development of emotions, cognitive processes, and behavior. Physiological dysregulation is a key transdiagnostic feature of psychopathology and target for multiple interventions. Alterations in arousal and reactivity are a core criterium for PTSD (American Psychiatric Association, 2013). There is debate if dysregulation of biological systems is a preexisting risk factor for the development of PTSD, a marker of PTSD itself, or consequence of disorder (Heim & Nemeroff, 2009; Kirsch et al., 2011; McFarlane et al., 1994; Zoladz & Diamond, 2013). While the DSM highlights increased arousal and reactivity (e.g., hypervigilance, exaggerated startle response), the present findings indicated that those with PTSS not only demonstrate increased ANS function during stress, but also decreased sympathetic function at rest, which is consistent with the view that there may be two meaningful PTSD subtypes (dissociative and re-experiencing/hyperaroused) (Lanius et al., 2010; Wolf et al., 2012). The findings also highlight additional consideration of low arousal symptoms in the assessment of PTSS in youth.

Further, relations between biological factors and symptoms are bidirectional and intertwined. Effects of medication on the SNS (e.g., blood thinners, norepinephrine blocking or deactivating agents, adrenergic receptor agonists) are thought to effectively treat arousal symptoms (e.g., nightmares, startle responses) and PTSD-related anxiety (Lipov & Kelzenberg, 2012). Further, interventions for PTSS have been shown to alter ANS function in adults. A systematic review found that trauma-focused cognitive behavioral therapy (TF-CBT) was associated with decreased HR and BP across various rest and stress tasks (Zantvoord, Diehle, & Lindauer, 2013), and two pilot studies demonstrated that incorporation of biofeedback in trauma interventions corresponded to faster PTSD symptom reduction (Polak, Witteveen, Denys, & Olff, 2015; Tan, Dao, Farmer, Sutherland, & Gevirtz, 2011. In youth, an attachment-based intervention delivered to child protective service-involved families was shown to improve ANS function during parent-child interactions (Tabachnick 2019), and a core TF-CBT component is relaxation skills to reduce physiological arousal (Cohen, Mannarino, & Deblinger, 2016). The present findings in conjunction with previous research highlight that a better understanding of ANS functioning in its relation to the conceptualization of PTSD and that psychophysiological symptoms are malleable targets for interventions in children.

Agenda for Future Research

The findings of the current meta-analysis and review are helpful in identifying directions for future research, including improved delineation and measurement of key constructs, research designs, trauma- and stressor-related considerations, and social or environmental factors.

Measurement of stressor characteristics.

The sample characteristics and types of stressors participants experienced varied substantially in included studies, ranging from extreme trauma to no exposure. Studies examining youth who experienced a single-event, non-interpersonal trauma may vary in their physiological reactivity and PTSD symptomology as compared to those reporting effects for youth exposed to chronic, interpersonal traumatic events. As reviewed above, the allostatic load model posits that wear and tear on the body accumulates with exposure to repeated or chronic stress (McEwen, 1998). It is hypothesized that youth exposed to single-instances of trauma would demonstrate a hyperreactive physiological profile, while those exposed to multiple and/or chronic trauma would demonstrate a hyporeactive profile. Patterns of hyperresponsivity and hyporesponsivity to stress should continue to be assessed considering trauma type, severity, and duration in future studies. In addition, the trauma type may influence the pattern of physiological dysregulation. While some argue that the type of trauma experienced might not influence the stress response significantly, as stress-response systems are not sensitive to specific types of experiences, and individual differences likely play a larger role (Smith & Pollak, 2020), the PNS in particular is embedded within the social engagement system (Porges, 2003a, 2007), and those exposed to interpersonal trauma are at the highest risk of developing PTSD (Alisic et al., 2014). It is possible that youth exposed to interpersonal trauma would display more dysregulated PNS activity.

Time since traumatic experience.

Time between physiology measurement and trauma exposure may moderate the relation between physiology and PTSS. In a meta-analysis of adult studies, Morris et al. (2016) found that increased HR measured soon after trauma corresponded to subsequent increased PTSS. Two meta-analyses have shown that elevated HR immediately post-injury is a consistent predictor of the development of PTSS after pediatric injury or trauma, as well as persisting PTSS in injured children (Alisic et al., 2014; Brosbe et al., 2011). The most consistent finding in a review of physiology and PTSD in children was the predictivity of HR immediately post-trauma, while other measures were inconclusive (Kirsch et al., 2011). In a review of SNS measures and PTSS, a positive relation was found for both adults and children in acute (< 1 month) and intermediate (1 – 11 months) time frames, but not for peritraumatic (< 24 hours) and enduring (12 – 24 months) PTSS (Morris & Rao, 2013).

The present study incorporated current models of psychopathology (Kotov, et al., 2017) and integrated both dimensional and categorical approaches to PTSS. While some studies have indicated that physiological reacitivity does not vary as a function of subthreshold versus full PTSD in veterans (Costanzo, et al., 2016; Keane, et al., 1998), understanding of biological correlates of PTSS would be improved by integrating a framework reflecting the series of stages in the progression of the disorder (McFarlane, Lawrence-Wood, Van Hooff, Malhi, & Yehuda, 2017). It is plausible that those with full PTSD demonstrate a different biological profile than those with subclinical symptoms. The present results demonstrated no significant effects when constrained to samples reporting differences between those with and without PTSD diagnosis, although fewer studies reported on samples of participants with a full PTSD diagnosis (k = 10) than studies reporting PTSS on a continuum (k = 28). The literature on depression has become large enough to examine differences between those with full diagnosis versus symptoms. For example, a meta-analysis including data from over 1,700,000 participants, found that gender differences were more pronounced in those with Major Depression as compared to those with depression symptoms (Salk, Hyde, & Abramson, 2017). As the research on the relations between ANS and PTSS continues to develop, researchers should assess if biological correlates are similar or different for those with clinical versus subclinical PTSS. Further, about half of adult PTSD cases resolve with or without treatment, while about half demonstrate a chronic course (Morina, Wicherts, Lobbrecht, & Priebe, 2014). McFarlane and colleagues (2017) highlight that the development and course of PTSD is essential to consider, and may also correspond to different biological subtypes. Longitudinal studies of the relation between ANS function and PTSS are ciritical for examining the interplay of biological effects associated with PTSD and can also lead to optimization of treatment based on the progression of PTSS in individuals.

Study methodology.

The present review revealed which ANS measures have been more thoroughly studied and which require additional research. The field is particularly skewed toward utilization of HR as the main general ANS measure, as the majority of the included studies assessing the ANS reported HR. Few studies assessed BP, heart period, IBI, or low frequency HRV across baseline and stress tasks. For PNS measures, almost all studies reported RSA. One study reported effects for high frequency HRV during baseline and stress tasks. For SNS measures, most studies reported effects for SCL, some studies reported SCR, and some studies reported PEP. The comparison of distinct autonomic measures could yield important information about specific psychophysiological changes in youth. While in this first review of the relations between ANS measures with PTSS it was not possible to assess specific comparisons across single measures quantitatively, this should be targeted in future research.

The majority (62%) of studies in the present analyses were cross-sectional, highlighting the need for additional longitudinal assessments of the relation between physiology and PTSS in youth. There were not enough longitudinal studies for ANS and SNS indices measured during rest/neutral tasks or SNS measures taken during stress tasks to calculate effects. These areas should be targeted for future research. Another consideration is how to best measure physiological change over time when stress responsivity often covaries with resting/baseline arousal levels (i.e., law of initial values) and how to address and interpret change scores (Hinnant et al., 2017). For example, physiological resting states tend to impose limits on an individual’s responsivity to challenge (i.e., floor and ceiling effects), which could confound interpretation of stress reactivity scores. This is particularly relevant, as the present findings indicate a small but significant negative relationship between SNS measures at rest and PTSS. Therefore, failing to account for resting physiology may be problematic.

In addition, physiological systems are developing over time. Researchers should continue to acknowledge factors both between and within individuals when assessing if autonomic functioning represents continuous or discontinuous development over time. For example, Hinnant et al. (2017) demonstrated that within-individual stability and reliability of RSA was high in both childhood and adolescence, while RSA change scores were much less reliable in childhood, but reliable in adolescence. The authors cautioned researchers in their interpretations of change over time in RSA change scores in childhood. Further, their data suggest that the most important changes in RSA may be at the individual versus aggregate level. Hinnant et al. (2017) posited that by adolescence, stress responsivity is less variable over time and less affected by environmental influences at the phenotypic level. This has implications for when intense or chronic stress in a child’s environment occurs and how autonomic physiology is measured.

Health factors and other stress systems should also be considered. First, health risk behaviors are potential confounds for results of autonomic functioning. For example, smoking, alcohol overuse, and sleep disturbance have been found to mediate the association between PTSS and autonomic measures in young adults (Dennis, et al., 2014). This needs to be assessed and/or controlled for in future studies, particularly as health risk behaviors and sleep patterns change with age. Second, other systems are implicated in neurobiological stress reponses (i.e., hypothalamic-pituitary-adrenal [HPA] axis). HPA axis activation occurs more slowly and has a longer duration than ANS changes. While these systems are mostly studied separately, it is important to note that both show differences between youth with posttraumatic stress symptoms and healthy controls (Schuurmans, Nijhof, Cima, Scholte, Popma, & Otten, 2021).

Social and environmental factors.

While there is evidence that targeting physiological symptoms for individuals with PTSD can be beneficial (Lipov & Kelzenberg, 2012; Polak et al., 2015; Tan et al., 2011; Zantvoord et al., 2013), it is also necessary for PTSD research to maintain emphasis on mitigating conflict and adversity and increasing social support and caregiver relationships versus focusing primarily on biological factors. Syme and Hagan (2020) argue that PTSS may be aversive yet adaptive responses to adversity and indicate social, not medical, problems requiring social solutions. While biological measures are a priority for research as outlined by Research Domain Criteria (i.e., RDoC), developmental, social, and environmental influences (and their interaction) should be prioritized in future research (Cuthbert, 2020). Future research should prioritize investigation of social and environmental protective factors for youth.

Limitations

While the present meta-analysis has several strengths and quantitatively consolidates literature on the associations of autonomic measures with PTSS in youth, there are several limitations. First, while the present analyses focused solely on PTSS, youth exposed to traumatic events are at risk for a multitude of mental health disorders (Felitti et al., 1998), and autonomic physiology has also been shown to correspond to other internalizing and externalizing psychopathology (Dietrich et al., 2007; El-Sheikh et al., 2001; Graziano & Derefinko, 2013; McLaughlin, Alves, et al., 2014). Second, PTSS levels varied across studies. The present findings include analyses of categorical (e.g., PTSD diagnosis vs. no PTSD), as well as sub-clinical and continuous measures. Third, the present meta-analysis combined ANS measures together at the system level (e.g., ANS, PNS, SNS). While this was deemed advantageous in order to maximize inclusivity of studies and compare meaningful patterns across facets of the ANS, individual indices were not quantitatively assessed. Each may demonstrate distinct relations with PTSS. Fourth, the present review included a broad age range in order to assess autonomic functioning in relation to PTSS across the first two decades of life, yet additional factors are likely confounded with age and development (e.g., trauma exposure). Lastly, health risk behaviors, we were unable to assess health risk behaviors in the present study, which can confound findings. It is encouraged that these limitations be targeted in future research.

Conclusion

The current meta-analysis and review assessed the relation between various physiological (ANS [i.e., combined PNS and SNS influence], primarily PNS, and primarily SNS) measures and symptoms of posttraumatic stress in youth. Small, significant relations were found for PTSS and ANS measures in stress tasks, and SNS measures during rest tasks. PNS measures were not significantly related to PTSS, yet showed evidence of varying with sample age. All significant results of stress tasks demonstrated significant heterogeneity. Included studies varied greatly in terms of sample trauma exposure types and methodology. While psychophysiological dysregulation is not necessary or sufficient for a diagnosis of PTSD, an improved understanding of the relation between physiological arousal and PTSS in youth can inform prevention and intervention.

Appendix A. ANS Measure Categorizations

ANS	PNS	SNS
HR	RSA	PEP
HP/IBI	HF-HRV	SCL
BP		SCR
HRV
LF-HRV

Note. Determinations for categorizations of physiological measurements (i.e., ANS, PNS, SNS) in the present review based on psychophysiological literature.

ANS = autonomic nervous system; PNS = parasympathetic nervous system; SNS = sympathetic nervous system; HR = heart rate (generally measured in beats per minute); HP = heart period (variation in beat-to-beat interval); IBI = interbeat interval (variation in beat-to-beat interval; time interval between heart beats); BP = blood pressure; RSA = respiratory sinus arrhythmia; HRV = heart rate variability (changes in inter-beat intervals over time; proposed to provide a more sensitive index of the autonomic stress response than mean HR; shown to have both PNS and SNS influence when reported at low frequency and low-frequency and high-frequency ratio measures); HF-HRV = high frequency HRV (while this is still under investigation and debated, considered a primarily PNS measure); HFBP = high frequency blood pressure; PEP = pre-ejection period; SCL = skin conductance level; SCR = skin conductance response

ANS = autonomic nervous system; BP = blood pressure; HP = heart period (variation in beat-to-beat interval); HR = heart rate (generally measured in beats per minute); HF-HRV = high frequency heart rate variability (while this is still under investigation and debated, considered a primarily PNS measure); HRV = heart rate variability (changes in inter-beat intervals over time; proposed to provide a more sensitive index of the autonomic stress response than mean HR; shown to have both PNS and SNS influence when reported at low frequency and low-frequency and high-frequency ratio measures); IBI = interbeat interval (variation in beat-to-beat interval; time interval between heart beats); LF-HRV = low frequency HRV; PEP = pre-ejection period; PNS = parasympathetic nervous system; RSA = respiratory sinus arrhythmia; SCL = skin conductance level; SCR = skin conductance response; SNS = sympathetic nervous system

Appendix B. Descriptions of studies included in the current meta-analysis and sample characteristics

Study	Country	n	Age M (Range)	% Female	% White	ANS Measure(s)	ANS Domains	ANS Task(s)	PTSD Measure(s)	Study Type
Barry et al. (2015)	US	147	19.02 (18 to 25)	52.5	74.5	RSA	PNS	BL rest	YSSC	CS
Bryant et al. (2007)	AU	76	9.87 (7 to 12)	34.12	--	HR	ANS	BL rest	UCLA	L
Busso et al. (2014)	US	44	16.72 (14 to 19)	65	45.5	PEP, RSA	PNS, SNS	Stress task	IES	L
Castelda (2006)	US	47	19.26 (--)	72.3	44.7	HR, SCL	ANS, SNS	BL rest; stress task	SIDES	CS
Cohen et al. (2020)	US	88	12.05 (--)	60.5	36.4	PEP, RSA	PNS, SNS	BL rest; stress task	UCLA	CS
Chen & Ghazali (2021)	MY	606	16.9 (14 to 19)	63.2	0	BP, HR	ANS	BL rest	PCL	CS
D’Andrea et al. (2013)	US	54	18.8 (--)	65	0	HR, SCL, SCR	ANS, SNS	BL rest; stress task	PCL	CS
De Young et al. (2007)	AU	101	10.8 (7 to 16)	35	--	HR	ANS	Stress recovery	ADIS	L
Fergus et al. (2011)	US	58	19.6 (18+)	100	81	HR, SCL	ANS, SNS	Stress task	DEQ	CS
Gray et al. (2018)	US	247	5.08 (3 to 6)	38	19	RSA	PNS	BL task	PAPA	CS
Haag et al. (2019)	UK	76	10.05 (6 to 13)	39.5	89.5	HR, LF-HRV, HF-HRV	ANS, PNS	BL task; stress tasks	UCLA	L
Hilberdink et al. (2021)	NL	49	14.5 (--)	45.9	100	HR	ANS	BL rest; stress tasks	CRIES	CS
Iffland et al. (2020)	DE	78	16.63 (15 to 20)	74.4	--	HR	ANS	Stress task	TSCC, UCLA	CS
Ivanov et al. (2011)	US	25	10.4 (8 to 12)	36	15	HR	ANS	BL rest; BL task; stress task	PTSRI	CS
Jones-Alexander et al. (2005)	US	21	12.7 (8 to 17)	47.5	--	BP, HR, SCL	ANS, SNS	BL rest; stress tasks	CPTSDI, PCL	CS
Kassam-Adams et al. (2005)	US	190	11.2 (8 to 17)	25	43	HR	ANS	Stress recovery	CAPS	L
Katz & Gurtovenko (2015)	US	75	9.33 (6 to 12)	51	57	RSA	PNS	BL task	CPSS	L
Kirsch et al. (2015)	DE	34	12.69 (6 to 17)	62	--	HR, RSA, SCL	ANS, PNS, SNS	BL rest; BL task; stress tasks	CAPS	CS
Lipschutz et al. (2017)	US	48	9.7 (7 to 13)	41.7	37.5	RSA	PNS	BL rest; stress task	CPSS	CS
MacArthur (2011)	US	61	15.7 (13 to 17)	44.3	31.1	HRV	ANS	BL rest; stress task	TSCC	CS
Marsac et al. (2017)	US	96	10.6 (8 to 13)	35.5	52.1	HR	ANS	Stress recovery	CPSS	L
Marx et al. (2005)	US	95	19.55 (undergrads)	100	52.6	HR	ANS	Stress task	PDS	CS
McLaughlin et al. (2016)	US	90	13.58 (6 to 18)	47.8	51.1	SCR	SNS	BL task	UCLA	CS
Mikolajewski et al. (2018) *	US	36	4.98 (3 to 6)	36	5.6	RSA	PNS	BL task; stress tasks	PAPA	CS
Mikolajewski et al. (2018) *	US	26	7.38 (4 to 9)	36	5.6	RSA	PNS	BL task; stress tasks	PAPA	L
Mischel (2015)	US	43	14.12 (10 to 17)	48.8	72.1	HR, SCL	ANS, SNS	Stress task	CAPS	CS
Motsan et al. (2021)	IL	76	11.6 (--)	59.2	--	RSA	PNS	BL rest	Zero-to-Three, DAWBA	L
Nixon et al. (2010)	AU	48	11.84 (7 to 17)	31	88	HR	ANS	Stress recovery	CPSS	L
Nugent et al. (2006)	US	66	13.21 (--)	31.7	79.3	HR	ANS	Stress recovery	CAPS	L
Olsson et al. (2008)	AU	79	10.81 (7 to 16)	33	--	HR	ANS	Stress recovery	CTSQ	L
Pfefferbaum et al. (2011)	US	17	18.1 (13 to 25)	41.2	88	BP, HR	ANS	BL rest; stress task	IES	CS
Rea (2007)	US	66	8.8 (7 to 11)	50	39.4	HR	ANS	BL rest; BL task	UCLA	CS
Saxe et al. (2006)	US	61	12.15 (6 to 18)	32.8	--	HR	ANS	Stress recovery	Child PTSD Reaction Index	L
Scheeringa et al. (2004)	US	62	-- (1 to 6)	--	--	HP/IBI	ANS	Stress task	PTSD Semi-Structured Interview, Observational Record for Infants and Young Children	CS
Schuurmans et al. (2021)	NL	77	14.8 (10 to 18)	46	--	PEP, RSA	PNS, SNS	BL task; stress tasks	CROPS, PROPS	CS
Shenk et al. (2012)	US	110	17 (14 to 19)	100	42	RSA	PNS	BL rest; stress tasks	CTI	CS
Shenk et al. (2014)	US	110	17 (14 to 19)	100	42	RSA	PNS	BL rest; stress tasks	CTI	L
Zatzick et al. (2006a)	US	97	15.9 (12 to 18)	33	72	HR	ANS	Stress recovery	UCLA	CS
Zatzick et al. (2006b)	US	108	15.9 (12 to 18)	32	72	HR	ANS	Stress recovery	UCLA	L

Note. AU = Australia; DE = Germany; IL = Israel; MY = Malaysia; NL = The Netherlands; UK = United Kingdom; US = United States of America; CS = cross-sectional sample; L = longitudinal sample; ANS = autonomic nervous system; PNS = parasympathetic nervous system; SNS = sympathetic nervous system; BP = blood pressure; HP = heart period; HR = heart rate; HRV = heart rate variability; HF-HRV = high frequency HRV; LF-HRV = low frequency HRV; IBI = interbeat interval; PEP = pre-ejection period; RSA = respiratory sinus arrhythmia; SCL = skin conductance level; SCR = skin conductance response; BL = baseline; ADIS = Anxiety Disorders Interview Schedule; CAPS = Clinician-Administered PTSD Scale; CPSS = Children’s Posttraumatic Stress Symptoms Scale; CPTSDI = Children’s PTSD Inventory; CRIES = Children Revised Impact of Event Scale; CROPS = Child Report of Posttraumatic Stress Symptoms; CTI = Comprehensive Trauma Interview; CTSQ = Child Trauma Screening Questionnaire; DAWBA = Developmental and Well-Being Assessment; DEQ = Distressing Events Questionnaire; IES = Impact of Events Scale; PAPA = Preschool-Aged Psychiatric Assessment; PCL = Posttraumatic Stress Disorder Checklist; PDS = Posttraumatic Diagnostic Scale; PROPS = Parent Report of Posttraumatic Stress Symptoms; PTSRI = Posttraumatic Stress Reaction Index; SIDES = Structured Interview for Disorders of Extreme Stress; TSCC = Trauma Symptom Checklist for Children; UCLA = UCLA Post-Traumatic Stress Disorder Reaction Index; YSSC = Youth Symptom Survey Checklist.

-- = data not reported in study

^*Reported cross sectional and longitudinal results within the same publication. As cross sectional and longitudinal were analyzed separately, data was entered separately.

Appendix C. Additional Study Information

Study	Sample/Stressor Characteristics	Study Task Information	PTSS Measure	PTSS Informant
Barry et al. (2015)	No stress or trauma specified	BL rest	Questionnaire	Youth self-report
Bryant et al. (2007)	Various trauma (traumatic fall, MVA or other traumatic injury)	BL rest	Interview	Youth report
Busso et al. (2014)	Living in Boston metro area at the time of terrorist attack	Stress task: TSST (Reactivity score: speech – BL)	Questionnaire	Youth self-report
Castelda (2006)	Childhood sexual or physical abuse	BL rest: relaxing audiotape Stress task: Auditory startle task	Questionnaire	Youth self-report
Chen & Ghazali (2021)	Majority (90%) of sample had at least one lifetime trauma exposure. Most were exposed to natural disaster (26.5%), transportation accident (22.9%), and near drowning (8.4%). 13.7% of participating reported having PTSD symptoms.	BL rest	Questionnaire	Youth self-report
Cohen et al. (2020)	No trauma specified. Sample described as “at-risk, racially diverse, low-income” and recruited via public health and flyers in particular school districts; median income ~ $44,000	BL rest Stress task: Parent-child conflict discussion (Reactivity score: task – BL)	Interview	Multi-informant
D’Andrea et al. (2013)	Prescreened for trauma history and some PTSD symptoms. All had trauma exposure. Various trauma (IPV, sexual abuse, family abuse, community violence, physical attack, combat, life threatening non-IPV, witnessing, loss, other). On average, participants reported 4 different types.	BL rest Stress task: Auditory startle task	Questionnaire	Youth self-report
De Young et al. (2007)	Admitted to ED for 24 hrs minimum following accidental physical injury; falls, road traffic accidents, sporting events, burns, dog attacks, other	Stress recovery	Interview	Caregiver report
Fergus et al. (2011)	Exposure to mass campus shooting All measures taken 8.8 w post-shooting	Stress tasks: (1) Expressive writing about shooting; (2) Reading own writing about shooting	Questionnaire	Youth self-report
Gray et al. (2018)	Various trauma (accidental injury, witness community assault, Hurricane Katrina, DV)	BL task: Neutral video	Interview	Caregiver report
Haag et al. (2019)	Acute trauma leading to ED attendance. Majority road traffic accident or accidental injury	BL task: simple writing task Stress tasks: (1) child trauma narrative; (2) parent-child trauma narrative	Interview	Youth report
Hilberdink et al. (2021)	Children of mothers with a high BMI or high anxiety symptoms during pregnancy or trauma-exposed children with no, mild, or high PTSD symptom levels. Trauma-exposed youth were exposed to death, or emotional, medical, or physical trauma, accidents, or other trauma.	BL rest Stress task: social stress task, speech preparation	Questionnaire	Youth self-report
Iffland et al. (2020)	PTSD group included youth reporting childhood sexual or physical abuse after age 3	Stress tasks: (1) reading physically threatening words; (2) reading socially threatening words (Reactivity score: reading – neutral word BL)	Multiple Questionnaire & Interview	Youth report
Ivanov et al. (2011)	Children with aggressive behavior indicated by school suspension due to physical fighting. All participants had ADHD dx. Varied from no trauma to moderate/severe trauma (DV, sexual abuse, loss of biological parent)	BL rest BL task: Nonviolent film Stress task: Violent film	Questionnaire	Multi-informant
Jones-Alexander et al. (2005)	MVA involvement	BL rest Stress tasks (1) Mental arithmetic; (2) MVA audiotape	Multiple Questionnaire & Interview	Multi-informant
Kassam-Adams et al. (2005)	Admitted to level I pediatric trauma center following traffic-related injury as pedestrian, passenger, or bicyclist	Stress recovery	Interview	Youth report
Katz & Gurtovenko (2015)	Children of mothers recruited at local DV agency as a part of a larger intervention study for survivors of IPV	BL task: Neutral story	Questionnaire	Youth self-report
Kirsch et al. (2015)	History of >1 traumatic events after age 3. Varied trauma (physical, sexual abuse, DV, accident, death, or multiple).	BL rest BL task: Neutral script (Reactivity score: neutral script – BL) Stress task: Trauma script (Reactivity score: trauma script – BL)	Interview	Youth report
Lipschutz et al. (2017)	History of ≥1 traumatic event, ≥4 PTSD symptoms and functional impairment. Varied trauma (natural disaster, DV, attack, sexual, threatened with weapon, accident, witness murder or dead body).	BL rest Stress task: Trauma script (Reactivity score: task – BL)	Questionnaire	Multi-informant
MacArthur (2011)	Normal population sample from public schools	BL rest Stress task: Mental arithmetic	Questionnaire	Youth self-report
Marsac et al. (2017)	Hospitalized for acute injury 2 w prior to assessment.	Stress recovery: First available reading from medical record (scene arrival, ED arrival, ED admission)	Questionnaire	Youth self-report
Marx et al. (2005)	Half of sample with history sexual victimization in adolescence or adulthood	Stress task: Date rape auditory stimulus (Reactivity score: task – BL)	Questionnaire	Youth self-report
McLaughlin et al. (2016)	No trauma specified. Recruited sample from neighborhoods with high violent crime, low SES, and agencies working with families experiencing violence	BL task: pre-conditioning phase of conditioning task	Interview	Youth report
Mikolajewski et al. (2018) *	History of ≥1 traumatic event after 36 mos of age; 61% witnessed or directed experienced Hurricane Katrina related trauma. Recruited from trauma center, battered women’s, and Head Start programs.	BL task: Neutral video Stress tasks: (1) distress video; (2) mom’s trauma memory	Interview	Caregiver
Mikolajewski et al. (2018) **	History of ≥1 traumatic event after 36 mos of age; 61% witnessed or directed experienced Hurricane Katrina related trauma. Recruited from trauma center, battered women’s, and Head Start programs.	BL task: Neutral video Stress tasks: (1) distress video; (2) mom’s trauma memory	Interview	Multi-informant
Mischel (2015)	All exposed to traumatic event. Trauma types were not specified.	Stress task: Personalized trauma script imagery (Reactivity score: trauma – neutral scripts)	Interview	Youth report
Motsan et al. (2021)	War-exposed group (i.e., residing in Israeli town located near Gaza border with thousands of frequent and unpredictable rocket attacks since 2000) and control group (i.e., residing in comparable towns in central Israel). Of war-exposed youth, 51% met criteria for PTSD at least once in childhood.	BL rest	Interview	Multi-informant
Nixon et al. (2010)	Varied trauma (MVA, falls, assaults, burns, accidental injuries). Recruited from ED or pediatric inpatient unit. Screened for no current child abuse. All received opiate meds during hospitalization.	Stress recovery: Measured at ER triage	Questionnaire	Youth self-report
Nugent et al. (2006)	Admitted to trauma center. Varied trauma (MVA, sport accident, assault, burn, dog bite, fall, lawn mower accident, accidental gun injury, swimming accident, falling tree injury)	Stress recovery: Averaged measurements from EMS transport and ED admission readings	Interview	Youth report
Olsson et al. (2008)	Hospitalization for >24h. Varied accidental injury (fall, bike accident, sporting accident, burns, MVA, animal attack, eye injury, other)	Stress recovery: Measurements at ED admission and 24h post-admission	Questionnaire	Youth self-report
Pfefferbaum et al. (2011)	Half of the sample included children of parents who experienced Oklahoma City bombing 7 y prior	BL rest Stress task: Interview about bombing (overall task values and reactivity score: task – rest)	Questionnaire	Youth self-report
Rea (2007)	Half of sample exposed to IPV or DV	BL rest BL task: Neutral video (Reactivity score: neutral – BL and neutral end – neutral start)	Interview	Caregiver
Saxe et al. (2006)	Varied trauma (burns, MVA, falls, assaults)	Stress recovery: Averaged measurements during hospitalization	Questionnaire	Youth self-report
Scheeringa et al. (2004)	Varied trauma (MVA, accidental injuries, DV, invasive procedures). Youth had ≥1 PTSD sx. Recruited from level 1 trauma center, battered women’s shelters, outpatient mental health program for violence-exposed kids, cancer program.	Stress task: Trauma recall (Reactivity score: trauma – BL)	Multiple	Multi-informant
Schuurmans et al. (2021)	Trauma-exposed youth in residential care due to severe psychiatric or behavioral problems, parental problems, and/or unsafe home environment; clinical scores on the CRIES. Controls recruited from the general community; no DSM disorders, or current/recent psychiatric treatment.	BL task: aquatic video Stress tasks: (1) TRIER-C; (2) SSST	Questionnaire	Multi-informant
Shenk et al. (2012) - CS	Youth with substantiated maltreatment (physical abuse/neglect, sexual abuse, or multiple) within 12 mos. Comparison group of non-maltreated youth recruited from primary care outpatient settings with general medical complaints of at-risk kids	BL rest Stress tasks: (1) performance task – affect recognition; (2) watching parent-child conflict video. Stress task values were combined.	Interview	Youth report
Shenk et al. (2014) - L	Youth with substantiated maltreatment (physical abuse/neglect, sexual abuse, or multiple) within 12 mos. Comparison group of non-maltreated youth recruited from primary care outpatient settings with general medical complaints of at-risk kids	BL rest Stress tasks: (1) performance task – affect recognition; (2) watching parent-child conflict video. Stress task values were combined.	Interview	Youth report
Zatzick et al. (2006a) - CS	Admitted to level 1 trauma center. Trauma included intentional (assault) or unintentional (MVA) injuries, sports injuries, stabbings, gunshot, burn, other	Stress recovery: Measured at ER triage	Interview	Youth report
Zatzick et al. (2006b) – L	Admitted to level 1 trauma center. Trauma included intentional (assault) or unintentional (MVA) injuries, sports injuries, stabbings, gunshot, burn, other	Stress recovery: Measured at ER triage	Interview	Youth report

Note. BL = baseline; CRIES = Children’s Revised Impact of Event Scale; DV = domestic violence; ED = emergency department; ER = emergency room; EMS = emergency medical services; MVA = motor vehicle accident; SSST = Sing-a-Song Stress Test; TSST = Trier Social Stress Task; TSST-C = Trier Social Stress Task, Child Version.

^*Cross sectional study details; this paper reports different types of results (e.g., cross-sectional, longitudinal) as well as different PTSD measurements (e.g., caregiver report only, multi-informant) within the same publication. Therefore, it is entered twice to reflect different data extracted from the same publication.

^**Longitudinal study details; this paper reports different types of results (e.g., cross-sectional, longitudinal) as well as different PTSD measurements (e.g., caregiver report only, multi-informant) within the same publication. Therefore, it is entered twice to reflect different data extracted from the same publication.