Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

A Longitudinal Examination of Heart-Rate and Heart Rate Variability as Risk Markers for Child Posttraumatic Stress Symptoms in an Acute Injury Sample

A Longitudinal Examination of Heart-Rate and Heart Rate Variability as Risk Markers for Child... Heart rate (HR) alterations in the immediate aftermath of trauma-exposure have been proposed to be potentially useful markers for child and adolescent posttraumatic stress disorder (PTSD). However, it is not yet clear if this holds true for measures taken more distal to the trauma, and no studies have investigated the predictive validity of more sensitive HR variability (HRV) indices. We recruited 76 parent-child pairs (child age 6 to 13 years) after the child experienced a traumatic event leading to presentation at a hospital emergency department. At 1-month post trauma (T1), HR recordings were obtained at rest, and while children verbally recounted their traumatic experience, both alone and together with a parent. Child post-traumatic stress symptoms (PTSS) were assessed concurrently (T1), and at 3 (T2) and 6-month (T3) follow-ups. We found that for T1, elevated mean HR during trauma narratives, but not at baseline, was positively associated with PTSS, with some evidence that HRV-indices were negatively cross-sectionally associated with PTSS. Furthermore, T1 HR indices predicted PTSS at T2 and partially at T3,althoughthese effectsdid notholdwhenT1PTSS were added to the model. Findings suggest that, consistent with the adult literature, HR indices in children may be a concurrent marker of higher PTSS and may be predictive of longer term distress. The findings encourage further investigations that track child HR and HRV in relation to PTSS over time after trauma, in order to examine how biological profiles evolve in those with persistent symptoms. . . . . . Keywords Longitudinal Child Adolescent Posttraumatic stress disorder Heart-rate Heart-rate variability Introduction Electronic supplementary material The online version of this article (https://doi.org/10.1007/s10802-019-00553-2) contains supplementary Meta-analytic reviews have found that between 10 and 20% material, which is available to authorized users. of children and young people (CYP) will develop PTSD after exposure to traumatic events (Alisic et al. 2014;Hiller * Sarah L. Halligan s.l.halligan@bath.ac.uk et al. 2016). The first 6 months post-trauma may be particu- larly important, as a period during which initial distress may Department of Psychology, University of Bath, Bath, UK remit spontaneously or become entrenched (Hiller et al. Department of Consultation-Liaison-Psychiatry and Psychosomatic 2016). Since PTSD is known to have serious and persistent Medicine, University Hospital Zurich, Zürich, Switzerland negative consequences for future adjustment (Salmon and Department for Clinical Psychology and Psychotherapy, University Bryant 2002; Bolton et al. 2000), it is important to establish of Saarland, Saarbrücken, Germany at the earliest possible stage which young people are at risk Faculty of Medicine and Health Sciences, University of East Anglia, of persistent symptoms following trauma exposure, and to Norwich, UK understand the processes via which risk is conferred. One Research Department of Clinical, Educational and Health way of achieving this is to identify biomarkers that are asso- Psychology, University College London, London, UK ciated with the onset and progression of the disorder Department of Psychology, University of Oxford, Oxford, UK (Beauchaine and Thayer 2015). Heart rate (HR) measures have been proposed to be Department of Psychiatry and Mental Health, University of Cape Town, Cape Town, South Africa potentially useful PTSD biomarkers, as they are relatively 1812 J Abnorm Child Psychol (2019) 47:1811–1820 easy to obtain (Mayeux 2004;Olssonetal. 2008). evidence has focused primarily on investigating mean HR. Mechanistically, it has been hypothesized that HR eleva- A number of cross-sectional studies have compared basal tions commonly found after trauma exposure may reflect HR between PTSD and non-PTSD samples of CYP, recruited the adrenergic response at the time of the event, as well as months to years after the original incident (for a review, see ongoing physiological stress reactions (Bryant 2006). As Kirsch et al. 2011). However, unlike the relatively consistent such, they are believed to influence key PTSD-relevant pattern of elevated basal autonomic activation found in adult processes, such as initial fear conditioning and trauma PTSD, there is little evidence of equivalent changes in CYP. memory consolidation, and to be related to ongoing Across studies examining a range of age groups and trauma symptoms including heightened distress in response to types, all but one (Scheeringa et al. 2004) found no difference trauma reminders and generalised hyperarousal. in resting HR between CYP with PTSD and trauma-exposed In adults, there is consistent evidence that elevated basal and non-exposed controls (Gray et al. 2018; Jones-Alexander HR and heightened HR reactivity, particularly to trauma et al. 2005; Kirsch et al. 2015). Although much of this re- cues, are concurrently associated with PTSD, suggesting search has been limited by modest sample sizes, one study heightened sympathetic activation (Buckley and Kaloupek of 247 very young children aged 3–6 years also found no basal 2001). In addition, elevated basal HR assessed shortly after HR elevations in association with PTSD (Gray et al. 2018). exposure to a range of traumatic events relatively consistent- Given the challenges of measuring PTSD in younger children, ly predicts higher PTSS up to 24 months later (meta-analytic replication of the latter result in older children is desirable. weighted r = 0.13) (Morris et al. 2016). Studies of adult Other studies have focused on HR reactivity, rather than PTSD have also examined HR variability (HRV), which basal HR, as a PTSD marker in CYP. Using a range of non- assesses changes in inter-beat intervals over time, and has trauma related provocations, no differences in HR reactivity been proposed to provide a more sensitive index of the were found between trauma-exposed young people with and autonomic stress response than mean HR (e.g. Appelhans without PTSD across diverse tasks (Grasso and Simons 2012; and Luecken 2006). Gray et al. 2018; Jones-Alexander et al. 2005; Saltzman et al. Generally, higher HRV is believed to reflect a wider reper- 2005; Scheeringa et al. 2004; MacMillan et al. 2009). toire of responses to stress, while lower variability may reflect However, research focused on trauma-specific provocations a diminished capacity to cope (Kim et al. 2018). Consequently, has yielded more mixed findings. Thus, the aforementioned reduced HRV has been associated with decreased self- study of children aged 3–6 years found a significantly stronger regulation and motivation for social interactions (Kemp and decline in respiratory sinus arrhythmia (RSA) during trauma Quintana 2013), with psychological disorders including anxi- recall in those who suffered from PTSD, as compared to ety and depression (Bassett 2016;Chalmerset al. 2014), and trauma-exposed and non-exposed controls (Gray et al. 2018). with long term consequences such as higher cardiovascular RSA is proposed to comprise a measure of parasympathetic disease and increased mortality (Thayer et al. 2010). One mea- tone that relates to emotional regulation (Beauchaine 2015). A sure which is considered to be a reliable and interpretable index second study of 124 pre-school children found higher HR re- of HRV is Fourier-derived high frequency band power (HFBP, activity to a trauma stimulus in trauma-exposed as compared to 0.15–0.4 Hz) (Billman 2013). HFBP is believed to reflect non-exposed children, but no differences between those with vagally-modulated functioning of the parasympathetic nervous and without PTSD (Scheeringa et al. 2004). In contrast, two system (PSN), with a higher HFBP indicating higher HRV, and small studies that used trauma scripts as a provocation found thus a more adaptive stress response (De Bellis and Putnam that HR reactivity in CYP aged 7–18 with PTSD did not differ 1994; Shaffer and Ginsberg 2017). A second, but less frequent- from either trauma exposed or non-exposed controls (Jones- ly studied measure of HRV, is low-frequency band power Alexander et al. 2005; Kirsch et al. 2015). (LFBP, 0.04–0.15 Hz). While its interpretability is still debated In addition to cross-sectional evidence, key longitudinal due to a mix of sympathetic and parasympathetic influences, as studies have looked at the predictive value of HR in relation basal LFBP is often reduced after trauma exposure, subsequent to PTSD in CYP (Kassam-Adams et al. 2005). There is rela- increases have been proposed to reflect the restoration of the tively consistent evidence that elevated basal HR obtained sympatho-vagal balance (Nagpal et al. 2013; Reyes del Paso within 24 h of emergency department (ED) admission follow- et al. 2013; Shaffer and Ginsberg 2017). A recent meta- ing trauma-exposure is a positive predictor of short and long- analysis showed that in adults with PTSD relative to controls, term PTSS in CYP, with a meta-analysis of six studies yield- there is consistent evidence for decreased basal HFBP, and for ing a small but reliable effect (weighted r = 0.18) (Alisic et al. small, but significant reductions in basal LFBP (Nagpal et al. 2011). However, there has been little longitudinal study of HR 2013), indicating HRV as a promising risk marker for PTSD. assessments taken more than 24 h after ED admission. One In contrast to robust evidence of persistent cardiovascular study found that HR indices assessed at hospital discharge alterations in association with PTSD in the adult field, studies after accidental trauma did not predict 6 month PTSS in of CYP have yielded less consistent findings. To date, the CYP (Nugent et al. 2006). Learning more about the predictive J Abnorm Child Psychol (2019) 47:1811–1820 1813 value of HR measured more distal to the trauma is potentially for full sample flow chart), there were no significant group important, as measures obtained early after trauma exposure differences in child age (p = 0.59) or gender (p = 0.66). could be confounded by factors such as pain and injury sever- However, those who were included in the final sample had ity, and may not be available for all CYP. significantly lower triage scores on presenting to the ED (M = In sum, while longitudinal studies indicate that elevated 1.78 vs. M = 2.55, t(246) = 5.02, p < 0.001) than those not HR immediately following injury can predict later PTSS, ev- included (see also Hiller et al. 2018). As a lower triage score idence of more persistent changes is limited. In contrast to the indicates that medical attention is needed more urgently, chil- adult literature, generally elevated autonomic activity has not dren with more severe incidents may be somewhat over- been found in CYP with PTSD, and evidence as to whether or represented in the current sample. not there is greater reactivity to trauma stimuli is mixed. In addition, HRV may provide a more sensitive marker of the autonomic stress response but has been little investigated in Measures CYP. In order to address gaps in the extant literature, we examined whether HR indices obtained 1 month after child Descriptive Measures Descriptive data on family background trauma exposure were associated with PTSS observed concur- and child trauma were obtained from parent interviews and rently, and whether they were predictive of persistent PTSS 3 ED records. To determine objective trauma severity, triage and 6 months later, in a sample of 76 children aged 6–13 years. ratings for urgency of care determined by hospital nurses were We examined both mean HR and HRV (HFBP, LFBP) indi- used, ranging from 1 = immediate care required to 4 = less ces, obtained at rest and while the children provided narratives urgent. of the traumatic experience a) alone and b) together with their parents. We hypothesized that child mean HR, both at rest and University of California PTSD Reaction Index (PTSD-RI), Child during the narratives, would positively predict PTSS at 1, 3 Report Child PTSD symptom-severity at 1, 3 and 6 months and 6 months, while HRV indices would be negatively asso- was assessed using the child-report of the PTSD- RI ciated. Furthermore, we expected HR measures taken during (Steinberg et al. 2004), a measure of PTSS for children aged the trauma narratives to be a stronger predictor of PTSD 6 years and older. Children rate the presence of 17 PTSD symptoms than baseline indices. symptoms over 27-items, according to DSM-IV criteria using a 5-point Likert scale ranging from 0 (BNone of the time^)to 4 (BMost of the time^). The PTSD-RI has been found to Methods differentiate reliably between individuals with and without trauma exposure (Steinberg et al. 2004), and has a high test- Participants rest reliability (Steinberg et al. 2013). Total symptom scores show good internal consistency (in the current study, Participants derived from a longitudinal study involving 132 Cronbach’s α =0.89). children and adolescents (6–13 years) and their primary carers, who were recruited following child exposure to an Trauma Narratives Children were asked to provide two narra- acute trauma leading to ED attendance (for a description of tives of their traumatic experience, always in the same order: the overall sample, see Hiller et al. 2018). From this sample, the first in the presence of the researcher only (child narrative) 110 (84%) CYP were offered the opportunity to undergo HR and the second together with their parent (joint narrative). For assessments (equipment was not available for the initial 22 both narratives, instructions asked participants (either the families). Of these, 19 (17%) did not want to participate in child only, or the parent and child) to begin just before the HR assessments, while for a further seven (6%), the assess- event occurred and include anything that they thought was ment was not performed due to situational restraints (e.g., the important. For the joint narrative, dyads were asked to de- family had limited time or the family environment was not scribe the event together, and the experimenter left the room conducive to accurate HR data collection). Eight recordings until called back by the family. Following their initial free (9.5%) showed too much noise as R-peaks were not clearly recall, participants were provided with some standard, basic identifiable. This led to a final sample of n = 76 participants prompts to elicit further recollections. These cues contained (56.6% of the overall sample) being included in the present questions about the child’s thoughts and feelings during and analyses. No significant differences were found in age (p = after the event, and were either administered by the experi- 0.48), sex (p = 0.38), objective trauma severity (p =. 68) and menter for the child narrative, or by the parents in the joint PTSD scores (T1: p = 0.70, T2; p =0.39, T3: p = 0.15) in the narrative. There was no time limit on either narrative, and the group who completed the HR task vs. those who were not tasks were audio-recorded. We included both narratives in assessed. Compared to the overall eligible sample described analyses in order to see whether consistent patterns would in Hiller et al. (2018)(n =341, see supplementary materials be yielded across them. 1814 J Abnorm Child Psychol (2019) 47:1811–1820 HR Assessments Electrocardiographic (ECG) data were sam- assessments took place in the family home, while the shorter pled at a rate of 600 Hz, using the Bio Radio 150 User Unit T2 assessments were conducted using postal or online ques- (Cleveland Medical Devices Inc., Cleveland, OH, USA) and tionnaires to reduce assessment burden, unless a face to face the software BBioCapture^ (Biopac-Systems Goleta, CA, assessment was required (e.g., families requested it, younger USA). Two 2 cm diameter disposable Ag/ AgCl electrodes child age). Parents and children were assessed separately by were placed on the child’s chest, beneath both collar bones, trained researchers. During the 2 h assessment at T1, parents approximately 6–7 in. apart, and a ground electrode was and children first completed demographic and trauma-related placed on the elbow of the non-dominant hand to allow reli- questionnaires. Electrodes for the HR recordings were at- able data collection despite the child potentially moving. tached prior to questionnaire completion, and a 3-min baseline ECG data were pre-processed using the Software recording was obtained while the child completed a simple BANSLAB^ version 2.6 (Blechert et al. 2016). ECG record- writing task. Subsequently, HR was recorded for the child ings were visually inspected for artefacts and ectopic beats, narrative task, an anagram task the parent and child completed and R-peak markers automatically set by the software were together (a non-trauma specific, mildly stress-inducing task), manually corrected where necessary. Datasets only displaying and the joint parent-child trauma narrative. During the record- high noise, due to equipment failures or electrode detachment, ings, the child was instructed to sit as still as possible in order were removed upon visual inspection (see above). to reduce noise in the recordings. At T2 and T3, parents and Furthermore, as HRVanalyses are relatively sensitive to noise children completed a range of questionnaires, including re- distortions (e.g. Bilchick and Berger 2006;Camm etal. 1996), ports on current child PTSD symptoms. the criterion was set that to allow for reliable analyses, a con- tinuous recording of more than 120 s had to be obtained for Data Analytic Strategy the baseline section, and of more than 300 s for each narrative, as the latter were on average substantially longer. This led to Data were analysed using SPSS Version 22. A square root the exclusion of 3 subjects at baseline, 1 for the child narrative transformation was applied to UCLA-RI scores to address and 4 for the joint narrative. Furthermore, if more than 3 s of skewness. Correlational analyses were performed to deter- consecutive noise were present, the recording was cut, as re- mine basic associations between variables. Subsequently, liable reconstruction would not have been possible. linear regression analyses were used to investigate wheth- Subsequently, the longest Breadable^ segment was retained. er HR indices (mean HR, LFBP and HFBP) predicted This affected 9 subjects at baseline (with on average 78% of child PTSD symptoms at T1, T2 and T3. Residualized the recording being retained), 10 for the child narrative (77% change scores from baseline HR to task HR were calcu- retained) and 20 for the joint narrative (74% retained for this lated for both narratives to control for pre-existing base- subset of participants). line HR differences. Shapiro-Wilk tests indicated these After pre-processing, mean HR was extracted, and non- change scores were normally distributed. Since HR indi- parametric fast Fourier Transformation, using a Welch ces maybeassociated witheachother,eachHRpredictor window-size of 30 s and natural log transformation, was ap- was entered into a separate regression model. Where T1 plied to extract indices of HFBP (0.15–4 Hz) and LFBP HR indices were found to predict symptoms at T2 or T3, (0.04–0.15). In a final step, one participant was excluded analyses were re-run controlling for T1 symptom severity, due to showing an unrealistically high mean HR, and five to determine whether HR has predictive power above and participants due to exhibiting HFBP and LFBP measures of beyond initial symptoms. Age, sex and triage were con- more than 3 standard deviations above the mean, as these sidered as potential control variables for all analyses. The measurements likely reflect recording errors or equipment Benjamini-Hochberg procedure (Benjamini and Hochberg malfunctions (Minassian et al. 2015). This led to a final sam- 1995) with a false discovery rate set to 5% was used to ple size of n = 70 for baseline analyses, n =74 for the child control to control for multiple testing. All findings narrative and n = 63 for the joint narrative. remained significant. Procedure Results The study was approved by the Research Ethics Committee of the University of Bath and the Oxford A NHS Research Ethics Descriptive Statistics Committee. Eligible families were recruited via four hospital EDs within the UK. Informed consent was obtained from all The sample comprised 76 children (46 male) aged 6 to individual participants included in the study. Assessments 13 years. The majority of children had experienced a road were completed at 3 time-points: 1 month post-trauma (T1), traffic accident or other accidental injury as their index trauma. and 3 months (T2) and 6 months (T3) later. T1 and T3 Accompanying parents were predominantly mothers, aged J Abnorm Child Psychol (2019) 47:1811–1820 1815 26–60 years. Detailed child and parental descriptive data are Cross-Sectional Analyses at T1 provided in Table 1. Child age, sex and objective trauma severity were explored For both narrative tasks, correlational analyses revealed as potential covariates. Results showed that age was negative- change in mean HR to be positively associated with T1 ly associated with PTSS at T1 and T3 (PTSD-RI T1: r = PTSS; similarly, changes in LFBP/HFBP were inversely as- −0.33, p < 0.001, PTSD-RI T3: r = −0.27, p <0.05), as well sociated with T1 PTSS (Table 3). We explored these associa- as with the following HR indices: baseline HR (r = −0.28, tions further, using linear regression analyses to examine p < 0.05), joint narrative HR (r = −0.27, p < 0.05) and joint whether HR and HRV measures were cross-sectionally asso- narrative LFBP (r = 0.23, p < 0.05). No significant associa- ciated with child PTSS at T1, controlling for child age. For the tions with either PTSD or HR measures were found for sex child narrative (Table 4), regression analyses found that, in or triage category. Only child age was therefore controlled for accordance with our hypotheses, residualized change scores in regression analyses. for meanHRwerepositivelyassociatedwithT1 PTSS (R change compared to a model including age only = 0.09, p = 0.008), whereas HFBP power was inversely associated with Associations Between HR Indices and PTSD T1 PTSS (R change = 0.09, p = 0.011). However, the effect of LFBP was not significant once age was controlled for (R Mean child scores on the main variables of analysis are change = 0.04, p = 0.102). For the joint narrative, mean HR 2 2 provided in Table 2, while correlations between the main positively (R change = 0.06, p = 0.045) and HFBP (R variables are presented in Table 3. Preliminary correlation- change = 0.18, p <0.001) and LFBP (R change = 0.12, p = al analyses revealed that none of the HR measures obtained 0.004) negatively predicted symptoms, with small to medium at baseline were associated with PTSS at any time point. effect sizes (see Table 5). Consequently, baseline HR as a predictor of PTSS was not explored further in regression analyses. For the child and joint narratives, both cross-sectional and longitudinal cor- Predicting PTSD Symptoms at T2 relations between mean HR/ HRV residualized change scores and PTSS were identified (see Table 3), which were Basic correlational analyses for T2 PTSS (i.e., 3 month follow- further tested in linear regression analyses, controlling for up) revealed that T1 mean HR was positively associated with child age. child/ adolescent PTSS, whereas LFBP/HFBP showed inverse Table 1 Descriptive information Demographic characteristics Statistic (N =76) Parent characteristics Age in years, M(SD) 39.82 (6.53) Proportion mothers, N(%) 70 (92.1%) Proportion married or cohabiting N(%) 51 (67.1%) Education Status, N(%): School until 18 years or younger 35 (47.3%) Further Education (Vocational Training/Diplomas,…) 16 (21.6%) Higher Education 23 (31.1%) Child characteristics Age in years M, (SD) 10.05 (0.22) Proportion male, N(%) 46 (60.5%) Ethnicity- Caucasian, N(%) 68 (89.5%) Triage Category, N(%): 1 (immediate attention required) 37 (48.7%) 2 (very urgent) 16 (21.1%) 3 (urgent) 13 (17.1%) 4 (less urgent) 10 (13.2%) Days in Hospital (Min-Max, M(SD)) 0–28, 2.80 (5.40) Days of school missed (Min-Max, M (SD))0–28, 5.15 (6.22) Proportion requiring ambulance/ helicopter, N(%): 54 (71.0%) 1816 J Abnorm Child Psychol (2019) 47:1811–1820 Table 2 Child mean scores on Child variable Mean SD 95% CI: lower bound; upper bound main outcome measures PTSD-RI- T1 18.76 12.54 15.90; 21.63 PTSD-RI- T2 15.44 12.82 12.19; 18.69 PTSD-RI- T3 13.91 12.59 10.86; 16.95 Baseline Mean HR 90.07 10.17 87.76; 92.37 Baseline HFBP 7.67 1.08 7.42; 7.91 Baseline LFBP 8.91 0.84 8.69; 9.11 Child Narrative Mean HR 92.00 10.57 89.52; 94.43 Child Narrative LFBP 7.85 0.86 7.66; 8.04 Child Narrative HFBP 9.32 0.78 9.16; 9.49 Joint Narrative Mean HR 89.25 9.00 87.01; 91.46 Joint Narrative LFBP 8.03 0.81 7.83; 8.22 Joint Narrative HFBP 9.61 0.64 9.45; 9.76 Note: PTSD-RI T2 scores are based on N = 59 and T3 scores on N = 68 due to participant dropout, all other scores are based on N =76 at T1 SD standard deviation, HR heart-rate, HFBP high frequency band power, LFBP low frequency band power, CI confidence interval, PTSD-RI posttraumatic stress disorder reaction index associations, for both the child and joint narratives (Table 3). where it showed a negative association. Similarly, for the joint Again, regression was used to probe these effects further. For narrative, residualized change scores for mean HR positively the child narrative, when performing regression analyses control- predicted PTSS (R change = 0.18, p = 0.003), and for HFBP 2 2 ling for age, residualized change scores for mean HR positively (R change = 0.19, p = 0.002) and LFBP (R change = 0.11, predicted T2 PTSS (R change = 0.11, p = 0.014), while HFBP p = 0.018) negatively predicted PTSS when controlling for age, 2 2 (R change = 0.10, p = 0.021) and LFBP (R change = 0.09, p = which was a non-significant predictor in all models (see Table 5). 0.023) were negatively associated (Table 4). Age was only a We next re-ran the models of T2 PTSS, this time including significant predictor of PTSS in the model including LFBP, T1 PTSS as an additional predictor, in order to see whether T1 Table 3 Associations between PTSD symptoms at T1, T2 and T3 and heart-rate indices at 1 month post-trauma 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. PTSD 1. T1 symptoms 1.00 ** 2. T2 symptoms 0.51 1.00 ** ** 3. T3 symptoms 0.50 0.67 1.00 Mean HR 4. Baseline 0.10 0.22 0.20 1.00 ** ** 5. Child narrative 0.33 0.36 0.24 0.06 1.00 * ** ** 6. Joint narrative 0.27 0.44 0.22 0.07 0.63 1.00 HFBP ** 7. Baseline 0.03 −0.19 −0.14 −0.70 0.20 0.14 1.00 * * * ** 8. Child Narrative −0.29 −0.32 −0.32 −0.23 −0.59 −0.21 0.06 1.00 ** ** * ** ** ** 9. Joint Narrative −0.35 −0.38 −0.33 −0.12 −0.39 −0.41 0.03 0.66 1.00 LFBP ** ** 10. Baseline 0.09 −0.02 −0.10 −0.49 0.12 0.14 0.74 0.07 −0.08 1.00 * * ** ** ** ** 11. Child Narrative −0.25 −0.37 −0.25 −0.32 −0.33 −0.24 0.16 0.64 0.55 −0.02 1.00 ** * ** ** ** ** ** ** ** 12. Joint Narrative −0.40 −0.35 −0.36 −0.40 −0.35 −0.37 0.23 0.48 0.70 −0.04 0.64 1.00 HR heart-rate, HFBP high frequency band power, LFBP low frequency band power. Heart-rate scores for the narratives are based on residualized change scores between baseline and the respective narrative * p <0.05, ** p <0.01 J Abnorm Child Psychol (2019) 47:1811–1820 1817 Table 4 Regression analyses: prediction of T1, T2 and T3 PTSD symptoms from age and heart-rate change indices for child narrative Model, predictors T1 (1 month post-trauma) T2 (3 month follow-up) T3 (6 months follow-up) 2 2 2 Model 1 R = 0.13, F(2,66) = 6.28** R = 0.10, F(2,52) = 4.07* R = 0.08, F(2,59) = 3.50* Age B = −0.22 (0.10), β = −0.26* B = −0.11 (0.11), β = −0.13 B = −0.26 (0.13), β = −0.25* Mean HR B = 0.47 (0.21), β=0.26* B = 0.58 (0.24), β=0.32* B = 0.36 (0.13), β = 0.17 2 2 2 Model 2 R = 0.15, F(2,65) = 7.20** R = 0.10, F(2,52) = 3.95* R = 0.12, F(2,59) = 5.01** Age B = −0.24 (0.10), β = −0.29* B = −0.15 (0.11), β = −0.17 B = −0.27 (0.12), β = −0.27* HFBP B = −0.51 (0.20), β = −0.29* B = −0.24 (0.10), β = −0.31* B = −0.56 (0.26), β = −0.26* 2 2 2 Model 3 R = 0.10, F(2,66) = 4.90** R = 0.09, F(2,52) = 3.81* R = 0.09, F(2,59) = 4.14* Age B = −0.23 (0.10), β = −0.28* B = −0.11 (0.11), β = −0.13* B = −0.25 (0.13), β = −0.24 LFBP B = −0.32 (0.19), β = −0.19 B = −0.53 (0.23), β = −0.31* B = −0.44 (0.25), β = −0.22 HR heart-rate, HFBP high frequency band power, LFBP low frequency band power. For all HR indices, change scores from baseline to narrative were used * p <0.05, ** p <0.01 HR indices had any longitudinal predictive effect above and Similarly, for the joint narrative, residualized change scores beyond initial distress. Following the addition of T1 PTSS were significantly negatively associated with child PTSS for 2 2 scores to the regression model (B =0.46, SE = 0.12, β = HFBP (R change = 0.15, p = 0.003) and LFBP (R change = 0.47, p = 0.001), mean HR assessed during the T1 joint narra- 0.10, p = 0.017) (see Table 5 for full statistics). When includ- tive continued to be a positive predictor of T2 PTSS (B =0.46, ing T1 PTSS scores in regression models, all effects were SE =0.20, β =0.29, p =0.003; R change = 0.07, p =0.030), rendered non-significant (βs ranged −0.13 to −0.09, all model R = 0.37, F (3,44) = 10.12, p < 0.001. However, all p > 0.25). Thus, for all models at step 2, T1 PTSS was the other predictive effects of child narrative and joint narrative only significant predictor of T3 outcomes (25.5% of variance HR indices were rendered non-significant by the inclusion of explained by itself). T1 PTSS in the model (βs ranged −0.22 to 0.20, all p >0.07). Predicting PTSD Symptoms at T3 Discussion Hierarchical regression models were only run for those T1 HR We examined whether mean HR and HRV measures taken indices showing significant correlations with child PTSS at T3 1 month after exposure to a traumatic event are reliable pre- (i.e., 6 month follow-up) (see Table 3). For the child narrative, dictors of child PTSS, both concurrently and at 3 and 6 months residualized change scores for HFBP significantly negatively follow-ups. We found that baseline HR at 1 month was not predicted T3 symptoms when entered in a regression model correlated with PTSS cross-sectionally or longitudinally. By controlling for age (R change = 0.07, p =0.037; see Table 4). contrast, indices of HR reactivity in response to trauma Table 5 Regression analyses: prediction of T1, T2 and T3 PTSD symptoms from age and heart-rate change indices for joint narrative Model, predictors T1 (1 month) T2 (3 months) T3 (6 months) 2 2 2 Model 1 R = 0.13, F(2,56) = 5.30** R = 0.19, F(2,45) = 14.65 R = 0.09, F(2,51) = 3.65* Age B = −0.20 (0.11), β = −0.24 B = −0.06 (0.11), β = −0.07 B = −0.22 (0.13), β = −0.24 Mean HR B = 0.46 (0.22), β =0.26* B = 0.70 (0.22), β =0.44 ** B = 0.38 (0.26), β = 0.20 2 2 2 Model 2 R = 0.19, F(2, 56) = 7.82** R = 0.20, F(2, 45) = 6.84** R = 0.18, F(2,51) = 6.76** Age B = −0.26 (0.10), β = −0.31* B = −0.18 (0.11), β = −0.23 B = −0.27 (0.12), β = −0.29* HFBP B = −0.62 (0.21), β = −0.35** B = −0.75 (0.22), β = −0.44** B = −0.67 (0.24), β = −0.35** 2 2 2 Model 3 R = 0.26, F(2, 56) = 11.29** R = 0.14, F(2, 45) = 4.76* R = 0.21, F(2,51) = 7.97** Age B = −0.18 (0.10), β = −0.21 B = −0.10 (0.11), β = −0.12 B = −0.19 (0.12), β = −0.20 LFBP B = −0.82 (0.21), β = −0.45** B = −0.65 (0.24), β = −0.37* B = −0.80 (0.25), β = −0.40** HR heart-rate, HFBP high frequency band power, LFBP low frequency band power. For all HR indices, change scores from baseline to narrative were used * p <0.05, ** p <0.01 1818 J Abnorm Child Psychol (2019) 47:1811–1820 provocation (either child only or joint narrative) showed a possible that the joint narrative resulted in higher emotional relatively reliable pattern of correlations with symptoms involvement of the child, and consequently a pattern of HR cross-sectionally, and longitudinally at 3 month follow-up, reactivity that more strongly associated with PTSS. However, with less robust prediction of 6 month outcomes. the child narrative was always conducted first, meaning that We found no evidence that baseline HR parameters mea- order effects may equally account for the pattern of findings. sured at 1 month post-trauma are associated with concurrent Overall, our observations challenge previous studies which levels of PTSS in CYP. This is consistent with previously found no differences in mean HR between CYP with PTSD reported null findings in relation to basal HR and child and trauma controls when listening to idiosyncratic trauma PTSS in children (Jones-Alexander et al. 2005; Kirsch et al. scripts (Jones-Alexander et al. 2005;Kirsch etal. 2015). The 2015), but is in contrast to the adult literature where basal current findings are consistent with the observations of one elevations in sympathetic activity have commonly been re- previous study of pre-school children, which found higher ported (Pole 2007). It is conceivable that mental health linked autonomic responding (stronger reduction in RSA) to trauma lifestyle factors (e.g., increase tobacco and alcohol use) par- cues in those with PTSD (Gray et al. 2018), and indicate that tially explain PTSD related changes in basal HR in the adult physiological reactivity to trauma cues in association with literature. Moreover, the adult field has disproportionately re- PTSD in CYP may show a profile more similar to that seen ported on veteran samples, where chronic stress and trauma in adults than previously thought. exposure are more typical (Pole 2007), whereas we focused on We also found evidence that indices of HRV are cross- children exposed to acute, single incident trauma. It will be sectionally associated with PTSS in young people with small important to conduct more long term studies of resting HR in to medium effect sizes, specifically HFBP as an indicator of relation to child PTSD, in order to examine whether changes the PNS response (De Bellis and Putnam 1994), and LFBP as emerge over time and could contribute to evidence of in- a putative indicator of SNS-PNS balance (Nagpal et al. 2013). creased risk of cardiovascular disease in association with ad- Relative reductions in HRV change scores from baseline dur- verse childhood experiences (Danese and McEwen 2012; ing the narratives at 1 month post-trauma were negatively Suglia et al. 2017). associated with PTSS. Lower HRV has been linked to a di- We also found no evidence that baseline HR as measured at minished self-regulation (Kim et al. 2018;Kempand 1 month post-trauma was predictive of PTSS 3 and 6 months Quintana 2013), which makes our findings consistent with later. This contrasts with evidence that mean HR shortly fol- an overall dysregulated ANS response to trauma cues. The lowing ED admission positively predicts long-term PTSS current study thus lays a promising basis for further explora- (Kassam-Adams et al. 2005; Alisicetal. 2011). The current tions of the role of HRV as a child/ adolescent PTSD risk- null findings should be treated cautiously, as our study was not marker. Importantly, while the results in the current study were well powered to detect longitudinal effects of the magnitude upheld when correcting for multiple testing using the previously reported in the literature (r = 0.18, a small effect, Benjamini Hochberg procedure, replication using larger sam- based on meta-analysis by Alisic et al. 2011). However, the ples and applying more stringent criteria to control for multi- greater elapsed time since trauma exposure is one likely ex- ple testing is needed. planation for the lack of prediction in the current study. Analyses of longitudinal associations between physiologi- Theoretical accounts of longitudinal associations between el- cal reactivity to each of our trauma narratives at T1 and PTSS evated post-trauma HR and later PTSS highlight the potential at 3- and 6-month follow-up, showed patterns of associations for adrenergic activation to augment memory consolidation/ that largely replicated cross-sectional findings, although ef- conditioning for trauma stimuli as the putative underlying fects were somewhat less robust for the longer follow-up. mechanism (Bryant 2006), but the same account cannot easily Despite these observations suggesting that altered physiolog- be applied to HR parameters 1 month following trauma. ical reactivity could be a marker for persistent distress, only Future studies that complete HR assessments longitudinally, HR reactivity during the joint narrative at T2 predicted later starting from ED presentation, can help to address such ques- PTSS (medium effect size) once T1 symptom levels were tions relating to underlying mechanisms, and can identify op- controlled for. This is not entirely surprising, as PTSS at 1- timal time-points at which risk marker assessments for PTSS month post-trauma are one of the strongest predictors of long- should be conducted (Olsson et al. 2008). term PTSD risk in CYP (Trickey et al. 2012). Furthermore, In contrast to null findings for baseline HR, when we ex- according to our cross-sectional analyses, HR indices and amined reactivity to trauma cues using two narrative tasks, PTSD symptom measures obtained at 1 month have a sub- relatively higher mean HR as compared to baseline was stantial amount of common variance, potentially leading to a cross-sectionally associated with child/adolescent PTSS at decreased predictive validity of HR indices when these two 1 month post trauma with small to medium effect sizes, for potential risk-markers are combined. Thus, co-varying for both narratives. Effects were somewhat stronger for the concurrent symptoms in trying to predict later distress is a Bparent-child^ than the Bchild only^ narrative, and it is particularly stringent test of causality. Nonetheless, based on J Abnorm Child Psychol (2019) 47:1811–1820 1819 the current findings it is possible that physiological reactivity Compliance with Ethical Standards to trauma cues is a marker for concurrent distress, rather than a Funding This research was funded byanEconomic andSocial causal factor in the aetiology of PTSD. Research Council (ESRC) grant awarded to SLH (ES/K006290/1). Our study has several strengths, including a longitudinal design, the incorporation of both baseline and trauma reactiv- Conflict of Interest None ity measures, and sensitive measurement of HR and HRV. Nonetheless, findings must be considered in the light of sev- Ethical Approval Ethical approval was obtained from the University of eral limitations. First, as the HR measurements were only Bath and Oxford A NHS Research Ethics Committees. introduced several months into the study, and a number of Open Access This article is distributed under the terms of the Creative recordings had to be excluded due to high noise, the sample Commons Attribution 4.0 International License (http:// size was only moderately large, potentially limiting our power creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give to detect long-term effects. Second, while the age range was appropriate credit to the original author(s) and the source, provide a link smaller than in many previous studies, the sample still com- to the Creative Commons license, and indicate if changes were made. prised mixed developmental stages. With mean resting HRs changing substantially over the course of development (Kassam-Adams et al. 2005), further studies are required to investigate age effects in more detail and derive reference References values before HR assessments can be established as routine clinical markers. Third, the current study did not take into Alisic, E., Jongmans, M. J., van Wesel, F., & Kleber, R. J. (2011). account child pre- and post-traumatic medical/physical health Building child trauma theory from longitudinal studies: a meta-anal- ysis. Clinical Psychology Review, 31(5), 736–747. status, which could have potentially confounded heart-rate Alisic, E., Zalta, A. K., Van Wesel, F., Larsen, S. E., Hafstad, G. S., measures. Fourth, CYP in the current sample had all experi- Hassanpour, K., et al. (2014). Rates of post-traumatic stress disorder enced single-incident traumas. This limits the generalizability in trauma-exposed children and adolescents: meta-analysis. The of current findings to other populations, such as children ex- British Journal of Psychiatry, 204(5), 335–340. Appelhans, B. M., & Luecken, L. J. (2006). Heart rate variability as an posed to repeated or prolonged traumas, or those living in index of regulated emotional responding. Review of General high-risk contexts, who might exhibit different patterns of Psychology, 10(3), 229–240. HR reactivity following trauma exposure (Cloitre et al. Bassett, D. (2016). A literature review of heart rate variability in depres- 2009). Relatedly, symptoms were generally low at T3, with sive and bipolar disorders. Australian and New Zealand Journal of Psychiatry, 50(6), 511–519. few children showing clinically significant PTSS levels, Beauchaine, T. P. (2015). Respiratory sinus arrhythmia: a transdiagnostic which may have contributed to the lack of strong predictive biomarker of emotion dysregulation and psychopathology. Current effects. It may be of benefit to replicate the study in a clinical Opinion in Psychology, 3,43–47. sample, investigating the association between HR indices and Beauchaine, T. P., & Thayer, J. F. (2015). Heart rate variability as a PTSD diagnoses, which was not possible here due to small transdiagnostic biomarker of psychopathology. International Journal of Psychophysiology, 98(2), 338–350. numbers meeting diagnostic criteria. Finally, it is important to Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: note that there was a slight over-representation of children a practical and powerful approach to multiple testing. Journal of the with more severe injuries in our sample, as compared to the Royal Statistical Society. Series B (Methodological), 57,289–300. overall eligible sample, with findings requiring confirmation Bilchick, K. C., & Berger, R. D. (2006). Heart rate variability. Journal of Cardiovascular Electrophysiology, 17(6), 691–694. in a sample that does not show this bias. Billman, G. E. (2013). The effect of heart rate on the heart rate variability In sum, the current study extended present research on HR response to autonomic interventions. Frontiers in Physiology, 4, 222. indices as a risk-marker for child and adolescent PTSD in sev- Blechert, J., Peyk, P., Liedlgruber, M., & Wilhelm, F. H. (2016). eral directions, including an investigation of the predictive val- ANSLAB: integrated multichannel peripheral biosignal processing in psychophysiological science. Behavior Research Methods, 48(4), ue of HRVassessments, which had not previously been studied 1528–1545. in this population. Extending our knowledge on this matter may Bolton, D., O’Ryan, D., Udwin, O., Boyle, S., & Yule, W. (2000). The ultimately lead to improved routine assessments, as well as the long-term psychological effects of a disaster experienced in adoles- targeted implementation of preventive and therapeutic mea- cence: II: general psychopathology. The Journal of Child Psychology and Psychiatry and Allied Disciplines, 41(4), 513–523. sures, with the aim of buffering adverse outcomes following Bryant, R. A. (2006). Longitudinal psychophysiological studies of heart child and adolescent trauma exposure as effectively as possible. rate: mediating effects and implications for treatment. Annals of the New York Academy of Sciences, 1071(1), 19–26. Buckley, T. C., & Kaloupek, D. G. (2001). A meta-analytic examination Acknowledgements We thank the families who volunteered their time to of basal cardiovascular activity in posttraumatic stress disorder. participate in this study; ED staff at Bristol Royal Hospital for Children, Psychosomatic Medicine, 63(4), 585–594. Royal United Hospital Bath, Great Western Hospital and Gloucestershire Camm, A., Malik, M., Bigger, J., Breithardt, G., Cerutti, S., Cohen, R., Hospital; and Mark Lyttle, Rosie McGuire, Soraya Safazadeh, Lara et al. (1996). Heart rate variability: standards of measurement, phys- Felder, and Theo Metcalfe for their assistance with this project. iological interpretation and clinical use. Task Force of the European 1820 J Abnorm Child Psychol (2019) 47:1811–1820 Society of Cardiology and the North American Society of Pacing Morris, M. C., Hellman, N., Abelson, J. L., & Rao, U. (2016). Cortisol, and Electrophysiology. Circulation, 93(5), 1043–1065. heart rate, and blood pressure as early markers of PTSD risk: a Chalmers, J. A., Quintana, D. S., Abbott, M. J., & Kemp, A. H. (2014). systematic review and meta-analysis. Clinical Psychology Review, Anxiety disorders are associated with reduced heart rate variability: 49,79–91. a meta-analysis. Frontiers in Psychiatry, 5,80. Nagpal, M., Gleichauf, K., & Ginsberg, J. (2013). Meta-analysis of heart Cloitre, M., Stolbach, B. C., Herman, J. L., Kolk, B. v. d., Pynoos, R., rate variability as a psychophysiological indicator of posttraumatic Wang, J., et al. (2009). A developmental approach to complex stress disorder. Journal of Trauma & Treatment, 3(2167–1222), PTSD: childhood and adult cumulative trauma as predictors of symptom complexity. Journal of Traumatic Stress, 22(5), 399–408. Nugent, N. R., Christopher, N. C., & Delahanty, D. L. (2006). Emergency Danese, A., & McEwen, B. S. (2012). Adverse childhood experiences, medical service and in-hospital vital signs as predictors of subse- allostasis, allostatic load, and age-related disease. Physiology & quent PTSD symptom severity in pediatric injury patients. Journal Behavior, 106(1), 29–39. of Child Psychology and Psychiatry, 47(9), 919–926. De Bellis, M. D., & Putnam, F. W. (1994). The psychobiology of child- Olsson, K. A., Kenardy, J. A., De Young, A. C., & Spence, S. H. (2008). hood maltreatment. Child and Adolescent Psychiatric Clinics of Predicting children’s post-traumatic stress symptoms following hos- North America, 3(4), 663–678. pitalization for accidental injury: combining the Child Trauma Grasso, D. J., & Simons, R. F. (2012). Electrophysiological responses to Screening Questionnaire and heart rate. Journal of Anxiety threat in youth with and without posttraumatic stress disorder. Disorders, 22(8), 1447–1453. Biological Psychology, 90(1), 88–96. Pole, N. (2007). The psychophysiology of posttraumatic stress disorder: a Gray, S. A., Lipschutz, R. S., & Scheeringa, M. S. (2018). Young chil- meta-analysis. Psychological Bulletin, 133(5), 725. dren’s physiological reactivity during memory recall: associations Reyesdel Paso,G.A., Langewitz,W., Mulder, L.J., VanRoon, A.,& with posttraumatic stress and parent physiological synchrony. Duschek, S. (2013). The utility of low frequency heart rate var- Journal of Abnormal Child Psychology, 46(4), 871–880. iability as an index of sympathetic cardiac tone: a review with Hiller, R. M., Meiser-Stedman, R., Fearon, P., Lobo, S., McKinnon, A., emphasis on a reanalysis of previous studies. Psychophysiology, Fraser, A., et al. (2016). Research review: changes in the prevalence 50(5), 477–487. and symptom severity of child post-traumatic stress disorder in the Salmon, K., & Bryant, R. A. (2002). Posttraumatic stress disorder in year following trauma–A meta-analytic study. Journal of Child children: the influence of developmental factors. Clinical Psychology and Psychiatry, 57(8), 884–898. Psychology Review, 22(2), 163–188. Hiller, R. M., Meiser-Stedman, R., Lobo, S., Creswell, C., Fearon, P., Saltzman, K. M., Holden, G. W., & Holahan, C. J. (2005). The psycho- Ehlers, A., et al. (2018). A longitudinal investigation of the role of biology of children exposed to marital violence. Journal of Clinical parental responses in predicting children’s post-traumatic distress. Child and Adolescent Psychology, 34(1), 129–139. Journal of Child Psychology and Psychiatry, 59(7), 781–789. Scheeringa, M. S., Zeanah, C. H., Myers, L., & Putnam, F. (2004). Heart Jones-Alexander, J., Blanchard, E. B., & Hickling, E. J. (2005). period and variability findings in preschool children with posttrau- Psychophysiological assessment of youthful motor vehicle accident sur- matic stress symptoms. Biological Psychiatry, 55(7), 685–691. vivors. Applied Psychophysiology and Biofeedback, 30(2), 115–123. Shaffer, F., & Ginsberg, J. (2017). An overview of heart rate variability Kassam-Adams, N., Garcia-España, J. F., Fein, J. A., & Winston, F. K. metrics and norms. Frontiers in Public Health, 5,258. (2005). Heart rate and posttraumatic stress in injured children. Archives of General Psychiatry, 62(3), 335–340. Steinberg, A. M., Brymer, M. J., Decker, K. B., & Pynoos, R. S. (2004). Kemp, A. H., & Quintana, D. S. (2013). The relationship between mental The University of California at Los Angeles post-traumatic stress and physical health: insights from the study of heart rate variability. disorder reaction index. Current Psychiatry Reports, 6(2), 96–100. International Journal of Psychophysiology, 89(3), 288–296. Steinberg, A. M., Brymer, M. J., Kim, S., Briggs, E. C., Ippen, C. G., Kim, H.-G., Cheon, E.-J., Bai, D.-S., Lee, Y. H., & Koo, B.-H. (2018). Ostrowski, S. A., et al. (2013). Psychometric properties of the Stress and Heart Rate Variability: a Meta-Analysis and Review of UCLA PTSD reaction index: part I. Journal of Traumatic Stress, the Literature. Psychiatry Investigation, 15(3), 235. 26(1), 1–9. Kirsch, V., Wilhelm, F. H., & Goldbeck, L. (2011). Psychophysiological Suglia, S. F., Koenen, K. C., Boynton-Jarrett, R., Chan, P. S., Clark, C. J., characteristics of PTSD in children and adolescents: a review of the Danese, A., et al. (2017). Childhood and adolescent adversity and literature. Journal of Traumatic Stress, 24(2), 146–154. cardiometabolic outcomes: a scientific statement from the American Kirsch, V., Wilhelm, F. H., & Goldbeck, L. (2015). Psychophysiological Heart Association. Circulation, CIR. 0000000000000536. characteristics of pediatric posttraumatic stress disorder during Thayer, J. F., Yamamoto, S. S., & Brosschot, J. F. (2010). The relationship script-driven traumatic imagery. European Journal of of autonomic imbalance, heart rate variability and cardiovascular Psychotraumatology, 6(1), 25,471. disease risk factors. International Journal of Cardiology, 141(2), MacMillan, H. L., Georgiades, K., Duku, E. K., Shea, A., Steiner, M., 122–131. Niec, A., et al. (2009). Cortisol response to stress in female youths Trickey, D., Siddaway, A. P., Meiser-Stedman, R., Serpell, L., & Field, A. exposed to childhood maltreatment: results of the youth mood pro- P. (2012). A meta-analysis of risk factors for post-traumatic stress ject. Biological Psychiatry, 66(1), 62–68. disorder in children and adolescents. Clinical Psychology Review, Mayeux, R. (2004). Biomarkers: potential uses and limitations. NeuroRx, 32(2), 122–138. 1(2), 182–188. Minassian, A., Maihofer, A. X., Baker, D. G., Nievergelt, C. M., Geyer, M. A., & Risbrough, V. B. (2015). Association of predeployment heart rate variability with risk of postdeployment posttraumatic Publisher’sNote Springer Nature remains neutral with regard to juris- stress disorder in active-duty marines. JAMA Psychiatry, 72(10), dictional claims in published maps and institutional affiliations. 979–986. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Abnormal Child Psychology Springer Journals

A Longitudinal Examination of Heart-Rate and Heart Rate Variability as Risk Markers for Child Posttraumatic Stress Symptoms in an Acute Injury Sample

Download PDF
10 pages

Loading next page...
 
/lp/springer-journals/a-longitudinal-examination-of-heart-rate-and-heart-rate-variability-as-dpttVYDuh0

References (49)

Publisher
Springer Journals
Copyright
Copyright © 2019 by The Author(s)
Subject
Psychology; Child and School Psychology; Neurosciences; Public Health
ISSN
0091-0627
eISSN
1573-2835
DOI
10.1007/s10802-019-00553-2
Publisher site
See Article on Publisher Site

Abstract

Heart rate (HR) alterations in the immediate aftermath of trauma-exposure have been proposed to be potentially useful markers for child and adolescent posttraumatic stress disorder (PTSD). However, it is not yet clear if this holds true for measures taken more distal to the trauma, and no studies have investigated the predictive validity of more sensitive HR variability (HRV) indices. We recruited 76 parent-child pairs (child age 6 to 13 years) after the child experienced a traumatic event leading to presentation at a hospital emergency department. At 1-month post trauma (T1), HR recordings were obtained at rest, and while children verbally recounted their traumatic experience, both alone and together with a parent. Child post-traumatic stress symptoms (PTSS) were assessed concurrently (T1), and at 3 (T2) and 6-month (T3) follow-ups. We found that for T1, elevated mean HR during trauma narratives, but not at baseline, was positively associated with PTSS, with some evidence that HRV-indices were negatively cross-sectionally associated with PTSS. Furthermore, T1 HR indices predicted PTSS at T2 and partially at T3,althoughthese effectsdid notholdwhenT1PTSS were added to the model. Findings suggest that, consistent with the adult literature, HR indices in children may be a concurrent marker of higher PTSS and may be predictive of longer term distress. The findings encourage further investigations that track child HR and HRV in relation to PTSS over time after trauma, in order to examine how biological profiles evolve in those with persistent symptoms. . . . . . Keywords Longitudinal Child Adolescent Posttraumatic stress disorder Heart-rate Heart-rate variability Introduction Electronic supplementary material The online version of this article (https://doi.org/10.1007/s10802-019-00553-2) contains supplementary Meta-analytic reviews have found that between 10 and 20% material, which is available to authorized users. of children and young people (CYP) will develop PTSD after exposure to traumatic events (Alisic et al. 2014;Hiller * Sarah L. Halligan s.l.halligan@bath.ac.uk et al. 2016). The first 6 months post-trauma may be particu- larly important, as a period during which initial distress may Department of Psychology, University of Bath, Bath, UK remit spontaneously or become entrenched (Hiller et al. Department of Consultation-Liaison-Psychiatry and Psychosomatic 2016). Since PTSD is known to have serious and persistent Medicine, University Hospital Zurich, Zürich, Switzerland negative consequences for future adjustment (Salmon and Department for Clinical Psychology and Psychotherapy, University Bryant 2002; Bolton et al. 2000), it is important to establish of Saarland, Saarbrücken, Germany at the earliest possible stage which young people are at risk Faculty of Medicine and Health Sciences, University of East Anglia, of persistent symptoms following trauma exposure, and to Norwich, UK understand the processes via which risk is conferred. One Research Department of Clinical, Educational and Health way of achieving this is to identify biomarkers that are asso- Psychology, University College London, London, UK ciated with the onset and progression of the disorder Department of Psychology, University of Oxford, Oxford, UK (Beauchaine and Thayer 2015). Heart rate (HR) measures have been proposed to be Department of Psychiatry and Mental Health, University of Cape Town, Cape Town, South Africa potentially useful PTSD biomarkers, as they are relatively 1812 J Abnorm Child Psychol (2019) 47:1811–1820 easy to obtain (Mayeux 2004;Olssonetal. 2008). evidence has focused primarily on investigating mean HR. Mechanistically, it has been hypothesized that HR eleva- A number of cross-sectional studies have compared basal tions commonly found after trauma exposure may reflect HR between PTSD and non-PTSD samples of CYP, recruited the adrenergic response at the time of the event, as well as months to years after the original incident (for a review, see ongoing physiological stress reactions (Bryant 2006). As Kirsch et al. 2011). However, unlike the relatively consistent such, they are believed to influence key PTSD-relevant pattern of elevated basal autonomic activation found in adult processes, such as initial fear conditioning and trauma PTSD, there is little evidence of equivalent changes in CYP. memory consolidation, and to be related to ongoing Across studies examining a range of age groups and trauma symptoms including heightened distress in response to types, all but one (Scheeringa et al. 2004) found no difference trauma reminders and generalised hyperarousal. in resting HR between CYP with PTSD and trauma-exposed In adults, there is consistent evidence that elevated basal and non-exposed controls (Gray et al. 2018; Jones-Alexander HR and heightened HR reactivity, particularly to trauma et al. 2005; Kirsch et al. 2015). Although much of this re- cues, are concurrently associated with PTSD, suggesting search has been limited by modest sample sizes, one study heightened sympathetic activation (Buckley and Kaloupek of 247 very young children aged 3–6 years also found no basal 2001). In addition, elevated basal HR assessed shortly after HR elevations in association with PTSD (Gray et al. 2018). exposure to a range of traumatic events relatively consistent- Given the challenges of measuring PTSD in younger children, ly predicts higher PTSS up to 24 months later (meta-analytic replication of the latter result in older children is desirable. weighted r = 0.13) (Morris et al. 2016). Studies of adult Other studies have focused on HR reactivity, rather than PTSD have also examined HR variability (HRV), which basal HR, as a PTSD marker in CYP. Using a range of non- assesses changes in inter-beat intervals over time, and has trauma related provocations, no differences in HR reactivity been proposed to provide a more sensitive index of the were found between trauma-exposed young people with and autonomic stress response than mean HR (e.g. Appelhans without PTSD across diverse tasks (Grasso and Simons 2012; and Luecken 2006). Gray et al. 2018; Jones-Alexander et al. 2005; Saltzman et al. Generally, higher HRV is believed to reflect a wider reper- 2005; Scheeringa et al. 2004; MacMillan et al. 2009). toire of responses to stress, while lower variability may reflect However, research focused on trauma-specific provocations a diminished capacity to cope (Kim et al. 2018). Consequently, has yielded more mixed findings. Thus, the aforementioned reduced HRV has been associated with decreased self- study of children aged 3–6 years found a significantly stronger regulation and motivation for social interactions (Kemp and decline in respiratory sinus arrhythmia (RSA) during trauma Quintana 2013), with psychological disorders including anxi- recall in those who suffered from PTSD, as compared to ety and depression (Bassett 2016;Chalmerset al. 2014), and trauma-exposed and non-exposed controls (Gray et al. 2018). with long term consequences such as higher cardiovascular RSA is proposed to comprise a measure of parasympathetic disease and increased mortality (Thayer et al. 2010). One mea- tone that relates to emotional regulation (Beauchaine 2015). A sure which is considered to be a reliable and interpretable index second study of 124 pre-school children found higher HR re- of HRV is Fourier-derived high frequency band power (HFBP, activity to a trauma stimulus in trauma-exposed as compared to 0.15–0.4 Hz) (Billman 2013). HFBP is believed to reflect non-exposed children, but no differences between those with vagally-modulated functioning of the parasympathetic nervous and without PTSD (Scheeringa et al. 2004). In contrast, two system (PSN), with a higher HFBP indicating higher HRV, and small studies that used trauma scripts as a provocation found thus a more adaptive stress response (De Bellis and Putnam that HR reactivity in CYP aged 7–18 with PTSD did not differ 1994; Shaffer and Ginsberg 2017). A second, but less frequent- from either trauma exposed or non-exposed controls (Jones- ly studied measure of HRV, is low-frequency band power Alexander et al. 2005; Kirsch et al. 2015). (LFBP, 0.04–0.15 Hz). While its interpretability is still debated In addition to cross-sectional evidence, key longitudinal due to a mix of sympathetic and parasympathetic influences, as studies have looked at the predictive value of HR in relation basal LFBP is often reduced after trauma exposure, subsequent to PTSD in CYP (Kassam-Adams et al. 2005). There is rela- increases have been proposed to reflect the restoration of the tively consistent evidence that elevated basal HR obtained sympatho-vagal balance (Nagpal et al. 2013; Reyes del Paso within 24 h of emergency department (ED) admission follow- et al. 2013; Shaffer and Ginsberg 2017). A recent meta- ing trauma-exposure is a positive predictor of short and long- analysis showed that in adults with PTSD relative to controls, term PTSS in CYP, with a meta-analysis of six studies yield- there is consistent evidence for decreased basal HFBP, and for ing a small but reliable effect (weighted r = 0.18) (Alisic et al. small, but significant reductions in basal LFBP (Nagpal et al. 2011). However, there has been little longitudinal study of HR 2013), indicating HRV as a promising risk marker for PTSD. assessments taken more than 24 h after ED admission. One In contrast to robust evidence of persistent cardiovascular study found that HR indices assessed at hospital discharge alterations in association with PTSD in the adult field, studies after accidental trauma did not predict 6 month PTSS in of CYP have yielded less consistent findings. To date, the CYP (Nugent et al. 2006). Learning more about the predictive J Abnorm Child Psychol (2019) 47:1811–1820 1813 value of HR measured more distal to the trauma is potentially for full sample flow chart), there were no significant group important, as measures obtained early after trauma exposure differences in child age (p = 0.59) or gender (p = 0.66). could be confounded by factors such as pain and injury sever- However, those who were included in the final sample had ity, and may not be available for all CYP. significantly lower triage scores on presenting to the ED (M = In sum, while longitudinal studies indicate that elevated 1.78 vs. M = 2.55, t(246) = 5.02, p < 0.001) than those not HR immediately following injury can predict later PTSS, ev- included (see also Hiller et al. 2018). As a lower triage score idence of more persistent changes is limited. In contrast to the indicates that medical attention is needed more urgently, chil- adult literature, generally elevated autonomic activity has not dren with more severe incidents may be somewhat over- been found in CYP with PTSD, and evidence as to whether or represented in the current sample. not there is greater reactivity to trauma stimuli is mixed. In addition, HRV may provide a more sensitive marker of the autonomic stress response but has been little investigated in Measures CYP. In order to address gaps in the extant literature, we examined whether HR indices obtained 1 month after child Descriptive Measures Descriptive data on family background trauma exposure were associated with PTSS observed concur- and child trauma were obtained from parent interviews and rently, and whether they were predictive of persistent PTSS 3 ED records. To determine objective trauma severity, triage and 6 months later, in a sample of 76 children aged 6–13 years. ratings for urgency of care determined by hospital nurses were We examined both mean HR and HRV (HFBP, LFBP) indi- used, ranging from 1 = immediate care required to 4 = less ces, obtained at rest and while the children provided narratives urgent. of the traumatic experience a) alone and b) together with their parents. We hypothesized that child mean HR, both at rest and University of California PTSD Reaction Index (PTSD-RI), Child during the narratives, would positively predict PTSS at 1, 3 Report Child PTSD symptom-severity at 1, 3 and 6 months and 6 months, while HRV indices would be negatively asso- was assessed using the child-report of the PTSD- RI ciated. Furthermore, we expected HR measures taken during (Steinberg et al. 2004), a measure of PTSS for children aged the trauma narratives to be a stronger predictor of PTSD 6 years and older. Children rate the presence of 17 PTSD symptoms than baseline indices. symptoms over 27-items, according to DSM-IV criteria using a 5-point Likert scale ranging from 0 (BNone of the time^)to 4 (BMost of the time^). The PTSD-RI has been found to Methods differentiate reliably between individuals with and without trauma exposure (Steinberg et al. 2004), and has a high test- Participants rest reliability (Steinberg et al. 2013). Total symptom scores show good internal consistency (in the current study, Participants derived from a longitudinal study involving 132 Cronbach’s α =0.89). children and adolescents (6–13 years) and their primary carers, who were recruited following child exposure to an Trauma Narratives Children were asked to provide two narra- acute trauma leading to ED attendance (for a description of tives of their traumatic experience, always in the same order: the overall sample, see Hiller et al. 2018). From this sample, the first in the presence of the researcher only (child narrative) 110 (84%) CYP were offered the opportunity to undergo HR and the second together with their parent (joint narrative). For assessments (equipment was not available for the initial 22 both narratives, instructions asked participants (either the families). Of these, 19 (17%) did not want to participate in child only, or the parent and child) to begin just before the HR assessments, while for a further seven (6%), the assess- event occurred and include anything that they thought was ment was not performed due to situational restraints (e.g., the important. For the joint narrative, dyads were asked to de- family had limited time or the family environment was not scribe the event together, and the experimenter left the room conducive to accurate HR data collection). Eight recordings until called back by the family. Following their initial free (9.5%) showed too much noise as R-peaks were not clearly recall, participants were provided with some standard, basic identifiable. This led to a final sample of n = 76 participants prompts to elicit further recollections. These cues contained (56.6% of the overall sample) being included in the present questions about the child’s thoughts and feelings during and analyses. No significant differences were found in age (p = after the event, and were either administered by the experi- 0.48), sex (p = 0.38), objective trauma severity (p =. 68) and menter for the child narrative, or by the parents in the joint PTSD scores (T1: p = 0.70, T2; p =0.39, T3: p = 0.15) in the narrative. There was no time limit on either narrative, and the group who completed the HR task vs. those who were not tasks were audio-recorded. We included both narratives in assessed. Compared to the overall eligible sample described analyses in order to see whether consistent patterns would in Hiller et al. (2018)(n =341, see supplementary materials be yielded across them. 1814 J Abnorm Child Psychol (2019) 47:1811–1820 HR Assessments Electrocardiographic (ECG) data were sam- assessments took place in the family home, while the shorter pled at a rate of 600 Hz, using the Bio Radio 150 User Unit T2 assessments were conducted using postal or online ques- (Cleveland Medical Devices Inc., Cleveland, OH, USA) and tionnaires to reduce assessment burden, unless a face to face the software BBioCapture^ (Biopac-Systems Goleta, CA, assessment was required (e.g., families requested it, younger USA). Two 2 cm diameter disposable Ag/ AgCl electrodes child age). Parents and children were assessed separately by were placed on the child’s chest, beneath both collar bones, trained researchers. During the 2 h assessment at T1, parents approximately 6–7 in. apart, and a ground electrode was and children first completed demographic and trauma-related placed on the elbow of the non-dominant hand to allow reli- questionnaires. Electrodes for the HR recordings were at- able data collection despite the child potentially moving. tached prior to questionnaire completion, and a 3-min baseline ECG data were pre-processed using the Software recording was obtained while the child completed a simple BANSLAB^ version 2.6 (Blechert et al. 2016). ECG record- writing task. Subsequently, HR was recorded for the child ings were visually inspected for artefacts and ectopic beats, narrative task, an anagram task the parent and child completed and R-peak markers automatically set by the software were together (a non-trauma specific, mildly stress-inducing task), manually corrected where necessary. Datasets only displaying and the joint parent-child trauma narrative. During the record- high noise, due to equipment failures or electrode detachment, ings, the child was instructed to sit as still as possible in order were removed upon visual inspection (see above). to reduce noise in the recordings. At T2 and T3, parents and Furthermore, as HRVanalyses are relatively sensitive to noise children completed a range of questionnaires, including re- distortions (e.g. Bilchick and Berger 2006;Camm etal. 1996), ports on current child PTSD symptoms. the criterion was set that to allow for reliable analyses, a con- tinuous recording of more than 120 s had to be obtained for Data Analytic Strategy the baseline section, and of more than 300 s for each narrative, as the latter were on average substantially longer. This led to Data were analysed using SPSS Version 22. A square root the exclusion of 3 subjects at baseline, 1 for the child narrative transformation was applied to UCLA-RI scores to address and 4 for the joint narrative. Furthermore, if more than 3 s of skewness. Correlational analyses were performed to deter- consecutive noise were present, the recording was cut, as re- mine basic associations between variables. Subsequently, liable reconstruction would not have been possible. linear regression analyses were used to investigate wheth- Subsequently, the longest Breadable^ segment was retained. er HR indices (mean HR, LFBP and HFBP) predicted This affected 9 subjects at baseline (with on average 78% of child PTSD symptoms at T1, T2 and T3. Residualized the recording being retained), 10 for the child narrative (77% change scores from baseline HR to task HR were calcu- retained) and 20 for the joint narrative (74% retained for this lated for both narratives to control for pre-existing base- subset of participants). line HR differences. Shapiro-Wilk tests indicated these After pre-processing, mean HR was extracted, and non- change scores were normally distributed. Since HR indi- parametric fast Fourier Transformation, using a Welch ces maybeassociated witheachother,eachHRpredictor window-size of 30 s and natural log transformation, was ap- was entered into a separate regression model. Where T1 plied to extract indices of HFBP (0.15–4 Hz) and LFBP HR indices were found to predict symptoms at T2 or T3, (0.04–0.15). In a final step, one participant was excluded analyses were re-run controlling for T1 symptom severity, due to showing an unrealistically high mean HR, and five to determine whether HR has predictive power above and participants due to exhibiting HFBP and LFBP measures of beyond initial symptoms. Age, sex and triage were con- more than 3 standard deviations above the mean, as these sidered as potential control variables for all analyses. The measurements likely reflect recording errors or equipment Benjamini-Hochberg procedure (Benjamini and Hochberg malfunctions (Minassian et al. 2015). This led to a final sam- 1995) with a false discovery rate set to 5% was used to ple size of n = 70 for baseline analyses, n =74 for the child control to control for multiple testing. All findings narrative and n = 63 for the joint narrative. remained significant. Procedure Results The study was approved by the Research Ethics Committee of the University of Bath and the Oxford A NHS Research Ethics Descriptive Statistics Committee. Eligible families were recruited via four hospital EDs within the UK. Informed consent was obtained from all The sample comprised 76 children (46 male) aged 6 to individual participants included in the study. Assessments 13 years. The majority of children had experienced a road were completed at 3 time-points: 1 month post-trauma (T1), traffic accident or other accidental injury as their index trauma. and 3 months (T2) and 6 months (T3) later. T1 and T3 Accompanying parents were predominantly mothers, aged J Abnorm Child Psychol (2019) 47:1811–1820 1815 26–60 years. Detailed child and parental descriptive data are Cross-Sectional Analyses at T1 provided in Table 1. Child age, sex and objective trauma severity were explored For both narrative tasks, correlational analyses revealed as potential covariates. Results showed that age was negative- change in mean HR to be positively associated with T1 ly associated with PTSS at T1 and T3 (PTSD-RI T1: r = PTSS; similarly, changes in LFBP/HFBP were inversely as- −0.33, p < 0.001, PTSD-RI T3: r = −0.27, p <0.05), as well sociated with T1 PTSS (Table 3). We explored these associa- as with the following HR indices: baseline HR (r = −0.28, tions further, using linear regression analyses to examine p < 0.05), joint narrative HR (r = −0.27, p < 0.05) and joint whether HR and HRV measures were cross-sectionally asso- narrative LFBP (r = 0.23, p < 0.05). No significant associa- ciated with child PTSS at T1, controlling for child age. For the tions with either PTSD or HR measures were found for sex child narrative (Table 4), regression analyses found that, in or triage category. Only child age was therefore controlled for accordance with our hypotheses, residualized change scores in regression analyses. for meanHRwerepositivelyassociatedwithT1 PTSS (R change compared to a model including age only = 0.09, p = 0.008), whereas HFBP power was inversely associated with Associations Between HR Indices and PTSD T1 PTSS (R change = 0.09, p = 0.011). However, the effect of LFBP was not significant once age was controlled for (R Mean child scores on the main variables of analysis are change = 0.04, p = 0.102). For the joint narrative, mean HR 2 2 provided in Table 2, while correlations between the main positively (R change = 0.06, p = 0.045) and HFBP (R variables are presented in Table 3. Preliminary correlation- change = 0.18, p <0.001) and LFBP (R change = 0.12, p = al analyses revealed that none of the HR measures obtained 0.004) negatively predicted symptoms, with small to medium at baseline were associated with PTSS at any time point. effect sizes (see Table 5). Consequently, baseline HR as a predictor of PTSS was not explored further in regression analyses. For the child and joint narratives, both cross-sectional and longitudinal cor- Predicting PTSD Symptoms at T2 relations between mean HR/ HRV residualized change scores and PTSS were identified (see Table 3), which were Basic correlational analyses for T2 PTSS (i.e., 3 month follow- further tested in linear regression analyses, controlling for up) revealed that T1 mean HR was positively associated with child age. child/ adolescent PTSS, whereas LFBP/HFBP showed inverse Table 1 Descriptive information Demographic characteristics Statistic (N =76) Parent characteristics Age in years, M(SD) 39.82 (6.53) Proportion mothers, N(%) 70 (92.1%) Proportion married or cohabiting N(%) 51 (67.1%) Education Status, N(%): School until 18 years or younger 35 (47.3%) Further Education (Vocational Training/Diplomas,…) 16 (21.6%) Higher Education 23 (31.1%) Child characteristics Age in years M, (SD) 10.05 (0.22) Proportion male, N(%) 46 (60.5%) Ethnicity- Caucasian, N(%) 68 (89.5%) Triage Category, N(%): 1 (immediate attention required) 37 (48.7%) 2 (very urgent) 16 (21.1%) 3 (urgent) 13 (17.1%) 4 (less urgent) 10 (13.2%) Days in Hospital (Min-Max, M(SD)) 0–28, 2.80 (5.40) Days of school missed (Min-Max, M (SD))0–28, 5.15 (6.22) Proportion requiring ambulance/ helicopter, N(%): 54 (71.0%) 1816 J Abnorm Child Psychol (2019) 47:1811–1820 Table 2 Child mean scores on Child variable Mean SD 95% CI: lower bound; upper bound main outcome measures PTSD-RI- T1 18.76 12.54 15.90; 21.63 PTSD-RI- T2 15.44 12.82 12.19; 18.69 PTSD-RI- T3 13.91 12.59 10.86; 16.95 Baseline Mean HR 90.07 10.17 87.76; 92.37 Baseline HFBP 7.67 1.08 7.42; 7.91 Baseline LFBP 8.91 0.84 8.69; 9.11 Child Narrative Mean HR 92.00 10.57 89.52; 94.43 Child Narrative LFBP 7.85 0.86 7.66; 8.04 Child Narrative HFBP 9.32 0.78 9.16; 9.49 Joint Narrative Mean HR 89.25 9.00 87.01; 91.46 Joint Narrative LFBP 8.03 0.81 7.83; 8.22 Joint Narrative HFBP 9.61 0.64 9.45; 9.76 Note: PTSD-RI T2 scores are based on N = 59 and T3 scores on N = 68 due to participant dropout, all other scores are based on N =76 at T1 SD standard deviation, HR heart-rate, HFBP high frequency band power, LFBP low frequency band power, CI confidence interval, PTSD-RI posttraumatic stress disorder reaction index associations, for both the child and joint narratives (Table 3). where it showed a negative association. Similarly, for the joint Again, regression was used to probe these effects further. For narrative, residualized change scores for mean HR positively the child narrative, when performing regression analyses control- predicted PTSS (R change = 0.18, p = 0.003), and for HFBP 2 2 ling for age, residualized change scores for mean HR positively (R change = 0.19, p = 0.002) and LFBP (R change = 0.11, predicted T2 PTSS (R change = 0.11, p = 0.014), while HFBP p = 0.018) negatively predicted PTSS when controlling for age, 2 2 (R change = 0.10, p = 0.021) and LFBP (R change = 0.09, p = which was a non-significant predictor in all models (see Table 5). 0.023) were negatively associated (Table 4). Age was only a We next re-ran the models of T2 PTSS, this time including significant predictor of PTSS in the model including LFBP, T1 PTSS as an additional predictor, in order to see whether T1 Table 3 Associations between PTSD symptoms at T1, T2 and T3 and heart-rate indices at 1 month post-trauma 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. PTSD 1. T1 symptoms 1.00 ** 2. T2 symptoms 0.51 1.00 ** ** 3. T3 symptoms 0.50 0.67 1.00 Mean HR 4. Baseline 0.10 0.22 0.20 1.00 ** ** 5. Child narrative 0.33 0.36 0.24 0.06 1.00 * ** ** 6. Joint narrative 0.27 0.44 0.22 0.07 0.63 1.00 HFBP ** 7. Baseline 0.03 −0.19 −0.14 −0.70 0.20 0.14 1.00 * * * ** 8. Child Narrative −0.29 −0.32 −0.32 −0.23 −0.59 −0.21 0.06 1.00 ** ** * ** ** ** 9. Joint Narrative −0.35 −0.38 −0.33 −0.12 −0.39 −0.41 0.03 0.66 1.00 LFBP ** ** 10. Baseline 0.09 −0.02 −0.10 −0.49 0.12 0.14 0.74 0.07 −0.08 1.00 * * ** ** ** ** 11. Child Narrative −0.25 −0.37 −0.25 −0.32 −0.33 −0.24 0.16 0.64 0.55 −0.02 1.00 ** * ** ** ** ** ** ** ** 12. Joint Narrative −0.40 −0.35 −0.36 −0.40 −0.35 −0.37 0.23 0.48 0.70 −0.04 0.64 1.00 HR heart-rate, HFBP high frequency band power, LFBP low frequency band power. Heart-rate scores for the narratives are based on residualized change scores between baseline and the respective narrative * p <0.05, ** p <0.01 J Abnorm Child Psychol (2019) 47:1811–1820 1817 Table 4 Regression analyses: prediction of T1, T2 and T3 PTSD symptoms from age and heart-rate change indices for child narrative Model, predictors T1 (1 month post-trauma) T2 (3 month follow-up) T3 (6 months follow-up) 2 2 2 Model 1 R = 0.13, F(2,66) = 6.28** R = 0.10, F(2,52) = 4.07* R = 0.08, F(2,59) = 3.50* Age B = −0.22 (0.10), β = −0.26* B = −0.11 (0.11), β = −0.13 B = −0.26 (0.13), β = −0.25* Mean HR B = 0.47 (0.21), β=0.26* B = 0.58 (0.24), β=0.32* B = 0.36 (0.13), β = 0.17 2 2 2 Model 2 R = 0.15, F(2,65) = 7.20** R = 0.10, F(2,52) = 3.95* R = 0.12, F(2,59) = 5.01** Age B = −0.24 (0.10), β = −0.29* B = −0.15 (0.11), β = −0.17 B = −0.27 (0.12), β = −0.27* HFBP B = −0.51 (0.20), β = −0.29* B = −0.24 (0.10), β = −0.31* B = −0.56 (0.26), β = −0.26* 2 2 2 Model 3 R = 0.10, F(2,66) = 4.90** R = 0.09, F(2,52) = 3.81* R = 0.09, F(2,59) = 4.14* Age B = −0.23 (0.10), β = −0.28* B = −0.11 (0.11), β = −0.13* B = −0.25 (0.13), β = −0.24 LFBP B = −0.32 (0.19), β = −0.19 B = −0.53 (0.23), β = −0.31* B = −0.44 (0.25), β = −0.22 HR heart-rate, HFBP high frequency band power, LFBP low frequency band power. For all HR indices, change scores from baseline to narrative were used * p <0.05, ** p <0.01 HR indices had any longitudinal predictive effect above and Similarly, for the joint narrative, residualized change scores beyond initial distress. Following the addition of T1 PTSS were significantly negatively associated with child PTSS for 2 2 scores to the regression model (B =0.46, SE = 0.12, β = HFBP (R change = 0.15, p = 0.003) and LFBP (R change = 0.47, p = 0.001), mean HR assessed during the T1 joint narra- 0.10, p = 0.017) (see Table 5 for full statistics). When includ- tive continued to be a positive predictor of T2 PTSS (B =0.46, ing T1 PTSS scores in regression models, all effects were SE =0.20, β =0.29, p =0.003; R change = 0.07, p =0.030), rendered non-significant (βs ranged −0.13 to −0.09, all model R = 0.37, F (3,44) = 10.12, p < 0.001. However, all p > 0.25). Thus, for all models at step 2, T1 PTSS was the other predictive effects of child narrative and joint narrative only significant predictor of T3 outcomes (25.5% of variance HR indices were rendered non-significant by the inclusion of explained by itself). T1 PTSS in the model (βs ranged −0.22 to 0.20, all p >0.07). Predicting PTSD Symptoms at T3 Discussion Hierarchical regression models were only run for those T1 HR We examined whether mean HR and HRV measures taken indices showing significant correlations with child PTSS at T3 1 month after exposure to a traumatic event are reliable pre- (i.e., 6 month follow-up) (see Table 3). For the child narrative, dictors of child PTSS, both concurrently and at 3 and 6 months residualized change scores for HFBP significantly negatively follow-ups. We found that baseline HR at 1 month was not predicted T3 symptoms when entered in a regression model correlated with PTSS cross-sectionally or longitudinally. By controlling for age (R change = 0.07, p =0.037; see Table 4). contrast, indices of HR reactivity in response to trauma Table 5 Regression analyses: prediction of T1, T2 and T3 PTSD symptoms from age and heart-rate change indices for joint narrative Model, predictors T1 (1 month) T2 (3 months) T3 (6 months) 2 2 2 Model 1 R = 0.13, F(2,56) = 5.30** R = 0.19, F(2,45) = 14.65 R = 0.09, F(2,51) = 3.65* Age B = −0.20 (0.11), β = −0.24 B = −0.06 (0.11), β = −0.07 B = −0.22 (0.13), β = −0.24 Mean HR B = 0.46 (0.22), β =0.26* B = 0.70 (0.22), β =0.44 ** B = 0.38 (0.26), β = 0.20 2 2 2 Model 2 R = 0.19, F(2, 56) = 7.82** R = 0.20, F(2, 45) = 6.84** R = 0.18, F(2,51) = 6.76** Age B = −0.26 (0.10), β = −0.31* B = −0.18 (0.11), β = −0.23 B = −0.27 (0.12), β = −0.29* HFBP B = −0.62 (0.21), β = −0.35** B = −0.75 (0.22), β = −0.44** B = −0.67 (0.24), β = −0.35** 2 2 2 Model 3 R = 0.26, F(2, 56) = 11.29** R = 0.14, F(2, 45) = 4.76* R = 0.21, F(2,51) = 7.97** Age B = −0.18 (0.10), β = −0.21 B = −0.10 (0.11), β = −0.12 B = −0.19 (0.12), β = −0.20 LFBP B = −0.82 (0.21), β = −0.45** B = −0.65 (0.24), β = −0.37* B = −0.80 (0.25), β = −0.40** HR heart-rate, HFBP high frequency band power, LFBP low frequency band power. For all HR indices, change scores from baseline to narrative were used * p <0.05, ** p <0.01 1818 J Abnorm Child Psychol (2019) 47:1811–1820 provocation (either child only or joint narrative) showed a possible that the joint narrative resulted in higher emotional relatively reliable pattern of correlations with symptoms involvement of the child, and consequently a pattern of HR cross-sectionally, and longitudinally at 3 month follow-up, reactivity that more strongly associated with PTSS. However, with less robust prediction of 6 month outcomes. the child narrative was always conducted first, meaning that We found no evidence that baseline HR parameters mea- order effects may equally account for the pattern of findings. sured at 1 month post-trauma are associated with concurrent Overall, our observations challenge previous studies which levels of PTSS in CYP. This is consistent with previously found no differences in mean HR between CYP with PTSD reported null findings in relation to basal HR and child and trauma controls when listening to idiosyncratic trauma PTSS in children (Jones-Alexander et al. 2005; Kirsch et al. scripts (Jones-Alexander et al. 2005;Kirsch etal. 2015). The 2015), but is in contrast to the adult literature where basal current findings are consistent with the observations of one elevations in sympathetic activity have commonly been re- previous study of pre-school children, which found higher ported (Pole 2007). It is conceivable that mental health linked autonomic responding (stronger reduction in RSA) to trauma lifestyle factors (e.g., increase tobacco and alcohol use) par- cues in those with PTSD (Gray et al. 2018), and indicate that tially explain PTSD related changes in basal HR in the adult physiological reactivity to trauma cues in association with literature. Moreover, the adult field has disproportionately re- PTSD in CYP may show a profile more similar to that seen ported on veteran samples, where chronic stress and trauma in adults than previously thought. exposure are more typical (Pole 2007), whereas we focused on We also found evidence that indices of HRV are cross- children exposed to acute, single incident trauma. It will be sectionally associated with PTSS in young people with small important to conduct more long term studies of resting HR in to medium effect sizes, specifically HFBP as an indicator of relation to child PTSD, in order to examine whether changes the PNS response (De Bellis and Putnam 1994), and LFBP as emerge over time and could contribute to evidence of in- a putative indicator of SNS-PNS balance (Nagpal et al. 2013). creased risk of cardiovascular disease in association with ad- Relative reductions in HRV change scores from baseline dur- verse childhood experiences (Danese and McEwen 2012; ing the narratives at 1 month post-trauma were negatively Suglia et al. 2017). associated with PTSS. Lower HRV has been linked to a di- We also found no evidence that baseline HR as measured at minished self-regulation (Kim et al. 2018;Kempand 1 month post-trauma was predictive of PTSS 3 and 6 months Quintana 2013), which makes our findings consistent with later. This contrasts with evidence that mean HR shortly fol- an overall dysregulated ANS response to trauma cues. The lowing ED admission positively predicts long-term PTSS current study thus lays a promising basis for further explora- (Kassam-Adams et al. 2005; Alisicetal. 2011). The current tions of the role of HRV as a child/ adolescent PTSD risk- null findings should be treated cautiously, as our study was not marker. Importantly, while the results in the current study were well powered to detect longitudinal effects of the magnitude upheld when correcting for multiple testing using the previously reported in the literature (r = 0.18, a small effect, Benjamini Hochberg procedure, replication using larger sam- based on meta-analysis by Alisic et al. 2011). However, the ples and applying more stringent criteria to control for multi- greater elapsed time since trauma exposure is one likely ex- ple testing is needed. planation for the lack of prediction in the current study. Analyses of longitudinal associations between physiologi- Theoretical accounts of longitudinal associations between el- cal reactivity to each of our trauma narratives at T1 and PTSS evated post-trauma HR and later PTSS highlight the potential at 3- and 6-month follow-up, showed patterns of associations for adrenergic activation to augment memory consolidation/ that largely replicated cross-sectional findings, although ef- conditioning for trauma stimuli as the putative underlying fects were somewhat less robust for the longer follow-up. mechanism (Bryant 2006), but the same account cannot easily Despite these observations suggesting that altered physiolog- be applied to HR parameters 1 month following trauma. ical reactivity could be a marker for persistent distress, only Future studies that complete HR assessments longitudinally, HR reactivity during the joint narrative at T2 predicted later starting from ED presentation, can help to address such ques- PTSS (medium effect size) once T1 symptom levels were tions relating to underlying mechanisms, and can identify op- controlled for. This is not entirely surprising, as PTSS at 1- timal time-points at which risk marker assessments for PTSS month post-trauma are one of the strongest predictors of long- should be conducted (Olsson et al. 2008). term PTSD risk in CYP (Trickey et al. 2012). Furthermore, In contrast to null findings for baseline HR, when we ex- according to our cross-sectional analyses, HR indices and amined reactivity to trauma cues using two narrative tasks, PTSD symptom measures obtained at 1 month have a sub- relatively higher mean HR as compared to baseline was stantial amount of common variance, potentially leading to a cross-sectionally associated with child/adolescent PTSS at decreased predictive validity of HR indices when these two 1 month post trauma with small to medium effect sizes, for potential risk-markers are combined. Thus, co-varying for both narratives. Effects were somewhat stronger for the concurrent symptoms in trying to predict later distress is a Bparent-child^ than the Bchild only^ narrative, and it is particularly stringent test of causality. Nonetheless, based on J Abnorm Child Psychol (2019) 47:1811–1820 1819 the current findings it is possible that physiological reactivity Compliance with Ethical Standards to trauma cues is a marker for concurrent distress, rather than a Funding This research was funded byanEconomic andSocial causal factor in the aetiology of PTSD. Research Council (ESRC) grant awarded to SLH (ES/K006290/1). Our study has several strengths, including a longitudinal design, the incorporation of both baseline and trauma reactiv- Conflict of Interest None ity measures, and sensitive measurement of HR and HRV. Nonetheless, findings must be considered in the light of sev- Ethical Approval Ethical approval was obtained from the University of eral limitations. First, as the HR measurements were only Bath and Oxford A NHS Research Ethics Committees. introduced several months into the study, and a number of Open Access This article is distributed under the terms of the Creative recordings had to be excluded due to high noise, the sample Commons Attribution 4.0 International License (http:// size was only moderately large, potentially limiting our power creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give to detect long-term effects. Second, while the age range was appropriate credit to the original author(s) and the source, provide a link smaller than in many previous studies, the sample still com- to the Creative Commons license, and indicate if changes were made. prised mixed developmental stages. With mean resting HRs changing substantially over the course of development (Kassam-Adams et al. 2005), further studies are required to investigate age effects in more detail and derive reference References values before HR assessments can be established as routine clinical markers. Third, the current study did not take into Alisic, E., Jongmans, M. J., van Wesel, F., & Kleber, R. J. (2011). account child pre- and post-traumatic medical/physical health Building child trauma theory from longitudinal studies: a meta-anal- ysis. Clinical Psychology Review, 31(5), 736–747. status, which could have potentially confounded heart-rate Alisic, E., Zalta, A. K., Van Wesel, F., Larsen, S. E., Hafstad, G. S., measures. Fourth, CYP in the current sample had all experi- Hassanpour, K., et al. (2014). Rates of post-traumatic stress disorder enced single-incident traumas. This limits the generalizability in trauma-exposed children and adolescents: meta-analysis. The of current findings to other populations, such as children ex- British Journal of Psychiatry, 204(5), 335–340. Appelhans, B. M., & Luecken, L. J. (2006). Heart rate variability as an posed to repeated or prolonged traumas, or those living in index of regulated emotional responding. Review of General high-risk contexts, who might exhibit different patterns of Psychology, 10(3), 229–240. HR reactivity following trauma exposure (Cloitre et al. Bassett, D. (2016). A literature review of heart rate variability in depres- 2009). Relatedly, symptoms were generally low at T3, with sive and bipolar disorders. Australian and New Zealand Journal of Psychiatry, 50(6), 511–519. few children showing clinically significant PTSS levels, Beauchaine, T. P. (2015). Respiratory sinus arrhythmia: a transdiagnostic which may have contributed to the lack of strong predictive biomarker of emotion dysregulation and psychopathology. Current effects. It may be of benefit to replicate the study in a clinical Opinion in Psychology, 3,43–47. sample, investigating the association between HR indices and Beauchaine, T. P., & Thayer, J. F. (2015). Heart rate variability as a PTSD diagnoses, which was not possible here due to small transdiagnostic biomarker of psychopathology. International Journal of Psychophysiology, 98(2), 338–350. numbers meeting diagnostic criteria. Finally, it is important to Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: note that there was a slight over-representation of children a practical and powerful approach to multiple testing. Journal of the with more severe injuries in our sample, as compared to the Royal Statistical Society. Series B (Methodological), 57,289–300. overall eligible sample, with findings requiring confirmation Bilchick, K. C., & Berger, R. D. (2006). Heart rate variability. Journal of Cardiovascular Electrophysiology, 17(6), 691–694. in a sample that does not show this bias. Billman, G. E. (2013). The effect of heart rate on the heart rate variability In sum, the current study extended present research on HR response to autonomic interventions. Frontiers in Physiology, 4, 222. indices as a risk-marker for child and adolescent PTSD in sev- Blechert, J., Peyk, P., Liedlgruber, M., & Wilhelm, F. H. (2016). eral directions, including an investigation of the predictive val- ANSLAB: integrated multichannel peripheral biosignal processing in psychophysiological science. Behavior Research Methods, 48(4), ue of HRVassessments, which had not previously been studied 1528–1545. in this population. Extending our knowledge on this matter may Bolton, D., O’Ryan, D., Udwin, O., Boyle, S., & Yule, W. (2000). The ultimately lead to improved routine assessments, as well as the long-term psychological effects of a disaster experienced in adoles- targeted implementation of preventive and therapeutic mea- cence: II: general psychopathology. The Journal of Child Psychology and Psychiatry and Allied Disciplines, 41(4), 513–523. sures, with the aim of buffering adverse outcomes following Bryant, R. A. (2006). Longitudinal psychophysiological studies of heart child and adolescent trauma exposure as effectively as possible. rate: mediating effects and implications for treatment. Annals of the New York Academy of Sciences, 1071(1), 19–26. Buckley, T. C., & Kaloupek, D. G. (2001). A meta-analytic examination Acknowledgements We thank the families who volunteered their time to of basal cardiovascular activity in posttraumatic stress disorder. participate in this study; ED staff at Bristol Royal Hospital for Children, Psychosomatic Medicine, 63(4), 585–594. Royal United Hospital Bath, Great Western Hospital and Gloucestershire Camm, A., Malik, M., Bigger, J., Breithardt, G., Cerutti, S., Cohen, R., Hospital; and Mark Lyttle, Rosie McGuire, Soraya Safazadeh, Lara et al. (1996). Heart rate variability: standards of measurement, phys- Felder, and Theo Metcalfe for their assistance with this project. iological interpretation and clinical use. Task Force of the European 1820 J Abnorm Child Psychol (2019) 47:1811–1820 Society of Cardiology and the North American Society of Pacing Morris, M. C., Hellman, N., Abelson, J. L., & Rao, U. (2016). Cortisol, and Electrophysiology. Circulation, 93(5), 1043–1065. heart rate, and blood pressure as early markers of PTSD risk: a Chalmers, J. A., Quintana, D. S., Abbott, M. J., & Kemp, A. H. (2014). systematic review and meta-analysis. Clinical Psychology Review, Anxiety disorders are associated with reduced heart rate variability: 49,79–91. a meta-analysis. Frontiers in Psychiatry, 5,80. Nagpal, M., Gleichauf, K., & Ginsberg, J. (2013). Meta-analysis of heart Cloitre, M., Stolbach, B. C., Herman, J. L., Kolk, B. v. d., Pynoos, R., rate variability as a psychophysiological indicator of posttraumatic Wang, J., et al. (2009). A developmental approach to complex stress disorder. Journal of Trauma & Treatment, 3(2167–1222), PTSD: childhood and adult cumulative trauma as predictors of symptom complexity. Journal of Traumatic Stress, 22(5), 399–408. Nugent, N. R., Christopher, N. C., & Delahanty, D. L. (2006). Emergency Danese, A., & McEwen, B. S. (2012). Adverse childhood experiences, medical service and in-hospital vital signs as predictors of subse- allostasis, allostatic load, and age-related disease. Physiology & quent PTSD symptom severity in pediatric injury patients. Journal Behavior, 106(1), 29–39. of Child Psychology and Psychiatry, 47(9), 919–926. De Bellis, M. D., & Putnam, F. W. (1994). The psychobiology of child- Olsson, K. A., Kenardy, J. A., De Young, A. C., & Spence, S. H. (2008). hood maltreatment. Child and Adolescent Psychiatric Clinics of Predicting children’s post-traumatic stress symptoms following hos- North America, 3(4), 663–678. pitalization for accidental injury: combining the Child Trauma Grasso, D. J., & Simons, R. F. (2012). Electrophysiological responses to Screening Questionnaire and heart rate. Journal of Anxiety threat in youth with and without posttraumatic stress disorder. Disorders, 22(8), 1447–1453. Biological Psychology, 90(1), 88–96. Pole, N. (2007). The psychophysiology of posttraumatic stress disorder: a Gray, S. A., Lipschutz, R. S., & Scheeringa, M. S. (2018). Young chil- meta-analysis. Psychological Bulletin, 133(5), 725. dren’s physiological reactivity during memory recall: associations Reyesdel Paso,G.A., Langewitz,W., Mulder, L.J., VanRoon, A.,& with posttraumatic stress and parent physiological synchrony. Duschek, S. (2013). The utility of low frequency heart rate var- Journal of Abnormal Child Psychology, 46(4), 871–880. iability as an index of sympathetic cardiac tone: a review with Hiller, R. M., Meiser-Stedman, R., Fearon, P., Lobo, S., McKinnon, A., emphasis on a reanalysis of previous studies. Psychophysiology, Fraser, A., et al. (2016). Research review: changes in the prevalence 50(5), 477–487. and symptom severity of child post-traumatic stress disorder in the Salmon, K., & Bryant, R. A. (2002). Posttraumatic stress disorder in year following trauma–A meta-analytic study. Journal of Child children: the influence of developmental factors. Clinical Psychology and Psychiatry, 57(8), 884–898. Psychology Review, 22(2), 163–188. Hiller, R. M., Meiser-Stedman, R., Lobo, S., Creswell, C., Fearon, P., Saltzman, K. M., Holden, G. W., & Holahan, C. J. (2005). The psycho- Ehlers, A., et al. (2018). A longitudinal investigation of the role of biology of children exposed to marital violence. Journal of Clinical parental responses in predicting children’s post-traumatic distress. Child and Adolescent Psychology, 34(1), 129–139. Journal of Child Psychology and Psychiatry, 59(7), 781–789. Scheeringa, M. S., Zeanah, C. H., Myers, L., & Putnam, F. (2004). Heart Jones-Alexander, J., Blanchard, E. B., & Hickling, E. J. (2005). period and variability findings in preschool children with posttrau- Psychophysiological assessment of youthful motor vehicle accident sur- matic stress symptoms. Biological Psychiatry, 55(7), 685–691. vivors. Applied Psychophysiology and Biofeedback, 30(2), 115–123. Shaffer, F., & Ginsberg, J. (2017). An overview of heart rate variability Kassam-Adams, N., Garcia-España, J. F., Fein, J. A., & Winston, F. K. metrics and norms. Frontiers in Public Health, 5,258. (2005). Heart rate and posttraumatic stress in injured children. Archives of General Psychiatry, 62(3), 335–340. Steinberg, A. M., Brymer, M. J., Decker, K. B., & Pynoos, R. S. (2004). Kemp, A. H., & Quintana, D. S. (2013). The relationship between mental The University of California at Los Angeles post-traumatic stress and physical health: insights from the study of heart rate variability. disorder reaction index. Current Psychiatry Reports, 6(2), 96–100. International Journal of Psychophysiology, 89(3), 288–296. Steinberg, A. M., Brymer, M. J., Kim, S., Briggs, E. C., Ippen, C. G., Kim, H.-G., Cheon, E.-J., Bai, D.-S., Lee, Y. H., & Koo, B.-H. (2018). Ostrowski, S. A., et al. (2013). Psychometric properties of the Stress and Heart Rate Variability: a Meta-Analysis and Review of UCLA PTSD reaction index: part I. Journal of Traumatic Stress, the Literature. Psychiatry Investigation, 15(3), 235. 26(1), 1–9. Kirsch, V., Wilhelm, F. H., & Goldbeck, L. (2011). Psychophysiological Suglia, S. F., Koenen, K. C., Boynton-Jarrett, R., Chan, P. S., Clark, C. J., characteristics of PTSD in children and adolescents: a review of the Danese, A., et al. (2017). Childhood and adolescent adversity and literature. Journal of Traumatic Stress, 24(2), 146–154. cardiometabolic outcomes: a scientific statement from the American Kirsch, V., Wilhelm, F. H., & Goldbeck, L. (2015). Psychophysiological Heart Association. Circulation, CIR. 0000000000000536. characteristics of pediatric posttraumatic stress disorder during Thayer, J. F., Yamamoto, S. S., & Brosschot, J. F. (2010). The relationship script-driven traumatic imagery. European Journal of of autonomic imbalance, heart rate variability and cardiovascular Psychotraumatology, 6(1), 25,471. disease risk factors. International Journal of Cardiology, 141(2), MacMillan, H. L., Georgiades, K., Duku, E. K., Shea, A., Steiner, M., 122–131. Niec, A., et al. (2009). Cortisol response to stress in female youths Trickey, D., Siddaway, A. P., Meiser-Stedman, R., Serpell, L., & Field, A. exposed to childhood maltreatment: results of the youth mood pro- P. (2012). A meta-analysis of risk factors for post-traumatic stress ject. Biological Psychiatry, 66(1), 62–68. disorder in children and adolescents. Clinical Psychology Review, Mayeux, R. (2004). Biomarkers: potential uses and limitations. NeuroRx, 32(2), 122–138. 1(2), 182–188. Minassian, A., Maihofer, A. X., Baker, D. G., Nievergelt, C. M., Geyer, M. A., & Risbrough, V. B. (2015). Association of predeployment heart rate variability with risk of postdeployment posttraumatic Publisher’sNote Springer Nature remains neutral with regard to juris- stress disorder in active-duty marines. JAMA Psychiatry, 72(10), dictional claims in published maps and institutional affiliations. 979–986.

Journal

Journal of Abnormal Child Psychology – Springer Journals

Published: May 10, 2019

There are no references for this article.