Temporal analysis of heart rate variability as a predictor of post traumatic stress disorder in road traffic accidents survivors

at SciVerse ScienceDirect

Journal of Psychiatric Research 46 (2012) 790e796

Contents lists available

Journal of Psychiatric Research

journal homepage: www.elsevier .com/locate/psychires

Temporal analysis of heart rate variability as a predictor of post traumaticstress disorder in road traffic accidents survivors

Abeer Shaikh al arab a,*, Laurence Guédon-Moreau b, François Ducrocq c, Sylvie Molenda d,Stéphane Duhema, Julia Salleron e, Isabelle Chaudieu f, Dina Bert a, Christian Libersa a,Guillaume Vaiva d,g

aClinical Investigation Center, CIC 9301 Inserm-CHU, Lille, FrancebDepartment of Pharmacology, Lille University Hospital, Lille, Francec SAMU59, Lille University Hospital, Lille, FrancedDepartment of Psychiatry, Lille University Hospital, Lille, FranceeDepartment of Biostatistic, Lille University Hospital, Lille, Francef Inserm U1061, Montpellier F-34093, Franceg Laboratoire de Neurosciences Fonctionnelles et Pathologies (CNRS), Université Lille Nord de France, France

a r t i c l e i n f o

Article history:Received 24 May 2011Received in revised form16 December 2011Accepted 17 February 2012

Keywords:Heart rate variabilityPost traumatic stress disorderRoad traffic accidentRisk factor

* Corresponding author. Hôpital Cardiologique, CHgation clinique, Bd du Pr Jules Leclercq, 59037 Lille Ce68 91; fax: þ33 3 20 44 68 90.

E-mail address: Abeer.SHAIKHALARAB@CHRU-LILL

0022-3956/$ e see front matter � 2012 Elsevier Ltd.doi:10.1016/j.jpsychires.2012.02.006

a b s t r a c t

Background: Road Traffic Accidents (RTA) are most probably the leading cause of post traumatic stressdisorder (PTSD) in developed countries. The autonomic nervous system (ANS) disturbances, due topsychological trauma, are part of the pathophysiology of PTSD. The aim of the present study was todetermine whether early heart rate variability (HRV) measurement, a biomarker of the ANS function,could act as a predictor of PTSD development after a RTA.Methods: We prospectively investigated 35 survivors of RTA with both physical injury and psychologicaltrauma. HRV data were obtained from 24-h Holter ECG monitoring, which was performed on the secondday after the accident. Time domain analysis was applied to the inter-beat (RR) interval time series tocalculate the various parameters of HRV. PTSD status was assessed 2 and 6 months after RTA.Results: There was a global diminution of HRV measurements in the PTSD group at both 2 and 6 months.The variability index was the best predictor of PTSD with the area under the receiveroperating curve fordiscriminating PTSD at 6 months at 0.92 (95% CI: 0.785; 1.046). A cut-off at 2.19% yielded a sensitivity of85.7% and a specificity of 81.8% for PTSD. Positive and negative predictive values were respectively 75%and 90%. However, initial heart rate (HR) data were relevant at 2 months but not at 6 months.Conclusion: RTA survivors exhibiting lower parasympathetic modulation of HR, indexed by temporalanalysis of HRV, are more susceptible to developing PTSD as a short and long-term outcome.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

The psychological trauma related to Road Traffic Accidents (RTA)is the cause of many acute and chronic psychopathological disor-ders, some of which can generate a serious disability. Post trau-matic stress disorder (PTSD) is one of the most common sequelae ofRTA in developed countries. Its prevalence in these countries in themonths following RTA is estimated between 20% and 45% amongdrivers and passengers (Blanchard and Hickling, 2004). PTSD could

RU de Lille, Centre d’investi-dex, France. Tel.: þ33 3 20 44

E.FR (A. Shaikh al arab).

All rights reserved.

have serious, far-reaching consequences for victims as well as forsociety as a whole, resulting in suicidal attempts in 27% of the cases(Blanchard and Hickling, 2004; Vaiva G et al., 2011). Accordingly,identifying victims who are at risk of developing PTSD in theaftermath of trauma would appear to be mandatory for earlyprevention and management, in order to reduce the frequency andintensity of PTSD. Among themultiple studies that targeted the fateof psycho-traumatised subjects, only a few clinical and biologicalfactors of vulnerability such as psychiatric scales, cortisol and GABAblood level have been proposed, which have also limited predictiveability and require time-consuming measures (Bryant, 2003;Delahanty et al., 2003; Vaiva et al., 2004).

The association between elevated basal heart rate (HR), themost prominent autonomic feature, and PTSD, has been well

A. Shaikh al arab et al. / Journal of Psychiatric Research 46 (2012) 790e796 791

documented (Buckley and Kaloupek, 2001). However, studiesinvestigating initial HR (measured within one week of traumaticinjury) as a predictor of subsequent PTSD raise a controversy(Bryant, 2006). While many current studies reported that initiallyelevated HR following trauma was associated with later PTSD(Bryant et al., 2008; Zatzick et al., 2005; Kassam et al., 2005), otherstudies did not find such a link (Buckley et al., 2004; Kuhn et al.,2006). Furthermore, a negative correlation between initial HRand chronic PTSD was reported (Blanchard et al., 2002), but thisstudy had some methodological limitations (retrospective, unrep-resentative sample). It is worth noting that whereas many largestudies tended to confirm the association between elevated HR andlater PTSD, there was much variability in the HR levels and subse-quent PTSD. In addition, a current study with a large sample sizeconcluded that the initial HR was a weak and non independentpredictor of PTSD. However, the prevalence of PTSD was low(Kraemer et al., 2008). Recently, a study with a large samplereported that only HR measured at the scene of RTA and not HRmeasured at hospital admission predicted later PTSD (Coronaset al., 2011).

Further studies have examined the relationship between PTSDand HR using indices of heart rate variability (HRV), a biomarker ofthe autonomic nervous system (ANS) function. The fluctuationsbetween consecutive normal inter-beat (RR) intervals providea “dynamic map” of the interaction between both branches of ANS(parasympathetic and sympathetic). HRV measurement may bea more revealing physiological assessment of PTSD than HR acti-vation. While measures of elevated HR habitually assess sympa-thetic arousal, HRV measures the relative influence and interactionof both branches of ANS on HR responsiveness. Furthermore,standardized investigations of HRV represent an economical, non-invasive procedure allowing not only a qualitative, but alsoa good semi-quantitative estimation of the ANS function (TaskForce, 1996). Besides, it is probable that decreased HRV indicatingautonomic dysfunction explains the negative cardiovascularoutcome in PTSD (Kubzansky et al., 2007).

It was previously reported that autonomic dysfunction due toincreased sympathetic and/or reduced vagal activity demonstratedby low HRV, was involved in a variety of psychiatric and physio-logical disorders, including depression (Kemp et al., 2010), anxiety(Miu et al., 2009), and insomnia (Bonnet and Arand, 1998), all ofwhich are known to be highly comorbid with PTSD. As regardsPTSD, many studies reported its association with low HRV by usingsymptom provocation cue protocols. These studies indicated thatlower basal and elicited respiratory sinus arrhythmia (RSA),a measure of parasympathetic cardiac modulation, were frequentlyassociated with PTSD diagnosis and severity. Certain studies foundthat participants with PTSD had lower baseline RSA amplitudecompared to normal control subjects (Cohen et al., 1997; Blechertet al., 2007). However, other studies showed that RSA amplitudewas only lower in the PTSD group during the trauma cue and not atbaseline (Sahar et al., 2001; Keary et al., 2009). These disparatefindings concerning RSA at baseline might be explained by twostudies that examined HRV and PTSD severity. When HRV assess-ment studies ranked participants with PTSD according to their HRVamplitude, they found that only those with the lowest indices hadan elevated HR (Hopper et al., 2006) and a more pronounced HRarousal (Sack et al., 2004). These results suggested that more severeautonomic dysregulation might influence elevated HR as well aslater PTSD.

Moreover, an increase in HRV has been associated witha decrease in PTSD symptoms in a few pilot intervention studies.Recent studies provided preliminary support for the efficacy of RSAbiofeedback (pacing breath rhythms at approximately six breaths-per-minute) in facilitating the increase in HRV amplitude and

improving the physiological and psychological health of individualswith PTSD (Zucker et al., 2009).

The present study hypothesized that: 1) RTA survivors withlower parasympathetic mediated HRV parameters would havea higher risk of PTSD compared to those with higher HRV; 2) PTSDseverity would be related to the magnitude of a decrease in para-sympathetic mediated HRV indices. To test these hypotheses, wemeasured temporal parameters of HRV soon after trauma exposurein a subset of RTA survivors of a prospective pilot study, whose aimwas to assess low post-trauma GABA plasma levels as a predictivefactor in the development of acute PTSD (Vaiva et al., 2004). Wefollowed up the patients to define the short and long-term (2, 6months) psychopathological outcome.

2. Materials and methods

2.1. Participants

Men and women aged 18 years and over who had life-threatening traffic accidents with physical injuries and whorequired hospitalization up to 72 h, were recruited from the Trau-matology Department of Lille University Hospital and DouaiHospital.We excluded patients with head traumawith initial loss ofconsciousness for more than 15 min (based on reports of para-medics and witnesses), patients with organic brain disease ordementia, patients receiving long-term treatment with benzodi-azepines or anticonvulsants, patients with a history of alcoholabuse or addiction and victims who were drunk at the time of theaccident. Specific exclusion criteria for the validity of HRVmeasurements included the following: treatment with medicationaffecting the central or peripheral ANS (i.e., psychiatric medication[hypnotics and anxiolytics], antihypertensive drugs [including b-blockers and calcium channel blockers]) which was a restrictivecriterion, illnesses known to be associated with a reduction in HRV(e.g., cardiac diseases, apoplexy, diabetic or alcoholic autonomicneuropathy) and difficulty with installation of a Holter device onthe anterior chest wall. The study received the approval of theregional ethics committee (CPP 02/30). Written informed consentwas obtained from all participants in the study.

2.2. Instruments

General characteristics of the subjects, the whole circumstancesof the accident and the Injury Severity Score calculation (ISS) wereassessed (Smith, 1990). The Peritraumatic Distress Inventory (PDI)was evaluated to determine whether the subject fulfilled theinclusion criteria of A1 and A2 of the DSM-IV for PTSD (Brunet et al.,2001). The Mini International Neuropsychiatric Interview items(M.I.N.I) for Diagnostic and Statistical Manual of Mental Disorders(DSM-IV) were used to assess the current and past psychopatho-logical state of each patient (Lecrubier et al., 1997). A validatedmodified version of the Clinician-Administered PTSD Scale (CAPS)completed by telephone was used (Blake et al., 1995; Aziz andKenford, 2004). The CAPS is widely considered to be the “goldstandard” in PTSD assessment. It is a 30-item scale that examinesthe frequency and severity of PTSD symptoms as defined by DSM-IV. In addition to these items, there were other tests to assess socialand occupational functioning, global PTSD symptom severity, andresponse validity. This evaluationwas performed at 2 and 6monthsfollowing the accident.

2.3. Analysis of heart rate variability

Twenty-four hour Holter ECG monitoring using a two channelsELA Medical record was undertaken at the patient’s bedside on day

3 lost to follow-up

6 ECG artefacts

21 analysed subjects

3 lost to follow-up

7 subjects PTSD+

Recontact (2 months)

Recontact (6 months) 11 subjects PTSD-

10 subjectsPTSD-7 subjects PTSD+ 4 subjects PTSD

sub-threshold

5 not meeting inclusion criteria

Subjects victims of road

accidents

N = 35

18 analysed subjects

Fig. 1. Flow chart of the study (PTSD, post traumatic stress disorder syndrome; PTSDþ,patients with PTSD syndrome; PTSD�, patients free of PTSD syndrome).

A. Shaikh al arab et al. / Journal of Psychiatric Research 46 (2012) 790e796792

2 post-trauma. A substantial part of long-term HRV measures wasattributed to the dayenight differences (circadian rhythm). Thus,the Holter recordings had to contain at least 18 h of analysable ECGdata which were registered during similar environmentalcircumstances.

Temporal HRV parameters were calculated, using specific soft-ware (ELA medical SYNETEC, version 1.21), according to previouslypublished guidelines and recommendations (Task Force, 1996).Automatic and manual editing of the inter-beat (RR) data wereperformed carefully to ensure correct identification and classifica-tion of every heart beat together with elimination of the potentialartefacts (if >10%, the Holter ECG was not retained). HRV analysiswas executed blindly for patient characteristics and a PTSDoutcome.

We calculated the mean HR (bpm) and the following indices ofHRV: SDNN (standard deviation of normal-to-normal intervals, inmilliseconds); SDANN (standard deviation of 5-min mean values ofnormal-to-normal intervals, in milliseconds); the variability index(percentage of RR differences (RR ¼ RR Diff (i) � RR (i � 1)); rMSSD(root-mean-square successive differences of normal-to-normalintervals, in milliseconds); and pNN50 (proportion of successivenormal-to-normal interval differences>50 ms, in percent). RMSSD,pNN50 and the variability index are considered to be reliableindices of a parasympathetic state (Task Force, 1996; Kleiger et al.,2005). Other time domain variables reflect a mixture of para-sympathetic, sympathetic and different physiological influences.

2.4. Statistical analysis

A non-parametric test was required due to the small number ofsubjects (n < 30). Data were presented as median [interquartilerange, IQR]. Statistical analysis was performed for per-protocolsubjects. Continuous variables of HRV measures (between diag-nostic groups) were compared using the Mann Whitney test.Categorical variables of clinical characteristics were assessed usingthe Fisher exact test. Spearman coefficient correlations wereapplied to define the parameters of HRV potentially correlated withthe CAPS. Double-tailed p < 0.05 was considered statisticallysignificant. All statistical analyses were performed using SPSS,version 15.0.

Table 1Demographic and clinical characteristics according to PTSD status, 2 and 6 monthsafter a traumatic injury.

2 months 6 months

PTSDþ(n ¼ 11)

PTSD�(n ¼ 10)

PTSDþ(n ¼ 7)

PTSD�(n ¼ 11)

Age (years) 23 [20; 34] 26 [20; 38] 29 [21; 40] 20 [20; 27]Sex ratio (M/F) 6/5 9/1 6/1 7/4Smoking 3 3 1 3History of road

accident4 8 5 6

History of trauma 4 5 4 4Ttt with b-blockers 1 0 0 1Ttt with narcotic

analgesics8 5 7 5*

Psychiatric Ttt(Yes/No)

8/3* 2/8* 6/1* 2/9*

Depression status(Yes/No)

2/9 0/10 1/6 0/11

Initial PDI scores 21 [17; 27] 21 [11; 25] 21 [15; 25] 24 [14; 26]CAPS scores 42 [28; 57] 9 [4; 14]*** 51 [36; 58] 11 [4; 12]***

PTSD, post traumatic stress disorder syndrome; PTSDþ, patients with PTSDsyndrome; PTSD�, patients free of PTSD syndrome; Ttt, treatment; PDI, Peri-traumatic Stress Inventory; CAPS, Clinician-Administered PTSD Scale. *p < 0, 05;***p < 0, 001 (for comparison between PTSDþ and PTSD� at each evaluation).

3. Results

Thirty five victims had Holter ECG recordings. Among them, 5victims did not meet inclusion criteria (3 were drunk at the time ofthe accident and 2 were treated with b-blockers or antihyperten-sive drugs). Six subjects with ECG recordings of bad quality weresubsequently excluded from analysis and 3 victims dropped out ateach PTSD evaluation, leaving 21 and 18 subjects to be included inthe statistical analysis at 2 and 6 months, respectively (Fig. 1). Theage of the victims ranged from 18 to 74 years. The Injury SeverityScore was elevated at 6 [2; 12]. Patients’ characteristics are shownin Table 1. Participants who were not included in our analysis(n ¼ 14), did not differ significantly from those who were followedup in terms of age, gender, initial HR or injury severity.

At the 2 month follow-up, the total number of 21 participantswas reassessed; 7 of them met full criteria for PTSD and 4 wereconsidered “sub-threshold” as they fulfilled the criterion of re-experiencing symptoms in addition to criteria of either avoidanceor heightened arousal symptoms (Fig. 1); These 11 patients weregrouped as positive for PTSD symptoms (PTSDþ) and had CAPSscores of 42 [28; 57], compared to 9 [4; 14] in the group withoutPTSD (PTSD�) (n¼ 10) [p< 0.0001], whowas defined as those whodid not meet criteria on any symptom cluster for PTSD diagnosis.

At the 6 month follow-up, 18 participants were evaluated again;7 of themmet full criteria for PTSD (no sub-threshold patients) andhad CAPS scores of 51 [36; 58], compared to 11 [4; 12] in the groupPTSD� (n ¼ 11) [p < 0.0001]. Among these patients with PTSD, 5had PTSD at 2 months.

Both groups (PTSDþ and PTSD�) were matched for age, gender,history of traumatic events and smoking at 2 and 6 months.Different parameters of HRV and HR are listed in Table 2. The shortterm evaluation (2 months) revealed that mean HR was higher andall the indices of HRV were significantly reduced in the PTSDþgroup versus the PTSD� group. Patients with PTSD had lowervalues for SDNN (p ¼ 0.02), SDANN (p ¼ 0.04), the variability index(p ¼ 0.04), rMSSD (p ¼ 0.01), and PNN50 (p ¼ 0.04), but had highervalues for mean HR (p ¼ 0.01), compared to patients without PTSD.The long-term evaluation (6 months) revealed no significantdifferences for mean HR. Nearly all the indices of HRV were

Table 2HRV measures in patients with and without PTSD diagnosed 2 and 6 months post-injury.

HRV parameters/24 h 2 months 6 months

PTSDþ (n ¼ 11) PTSD� (n ¼ 10) PTSDþ (n ¼ 7) PTSD� (n ¼ 11)

pNN50 (%) 1.59 [0.68; 5.67] 9.95 [2.67; 42.52]* 1.59 [1.16; 4.33] 12.91 [3.10; 40.39]*rMSSD (ms) 17.70 [16.94; 27.35] 49.50 [22.72; 93.46]** 18.80 [17.34; 27.74] 52.66 [23.72; 89.93]*Variability Index (%) 1.91 [1.49; 2.61] 2.84 [1.98; 7.04]* 1.71 [1.49; 2.18] 3.30 [2.20; 7.01]**SDANN (ms) 75.92 [43.08; 88.64] 93.57 [77.33; 112.43]* 81.32 [57.82; 91.27] 95.87 [76.72; 107.95]SDNN (ms) 86.14 [67.99; 102.67] 118.32 [92.89; 154.60]* 93.50 [78.21; 104.57] 125.08 [95.07; 144.81]*HR (bpm) 93.50 [77.70; 97.70] 72.90 [66.33; 80. 80]** 78.60 [77.70; 94.00] 76.10 [66.50; 90.80]

PTSD, post traumatic stress disorder syndrome; PTSDþ, patients with PTSD syndrome; PTSD�, patients free of PTSD syndrome; HRV, heart rate variability; pNN50, proportionof successive normal-to-normal interval differences >50; rMSSD, root-mean-square successive differences of normal-to-normal intervals; Variability Index, percentage of RRdifferences; SDANN, standard deviation of average 5-min of normal-to-normal intervals; SDNN, standard deviation of normal-to-normal intervals; ms, milliseconds; HR, meanheart rate; bpm, beats per minute. Data are presented as median (IQR). *p < 0, 05, **p ¼ 0, 01.

A. Shaikh al arab et al. / Journal of Psychiatric Research 46 (2012) 790e796 793

significantly lower in the PTSDþ group, except for SDANN whichwas also reduced but did not attain statistical significance. Patientswith PTSD had lower values for SDNN (p ¼ 0.03), the variabilityindex (p ¼ 0.004), rMSSD (p ¼ 0.03), and PNN50 (p ¼ 0.03) (Fig. 2).

There was a significant negative correlation between the CAPSglobal score (indicating PTSD severity) and vagally mediated HRVindices (rMSSD, PNN50, the variability index) at the short and long-term evaluations (Table 3). On the other hand, there was nosignificant correlation between mean HR and PTSD severity, asindexed by the CAPS.

We found a high degree of correlation among the variousparameters of HRV, which was to be expected as these parametersexamined the same data set. The variability index was the best

Fig. 2. Basal heart rate variability (HRV) parameters and post traumatic stress disorder (PTSDfree of PTSD syndrome; RMSSD, root-mean-square successive differences of normal-to-normpNN50, proportion of successive normal-to-normal interval differences >50; variability ind

predictor of PTSD at 6 months with the area under the receiver-operating curve for discriminating PTSD at 6 months being 0.92(95% CI: 0.785; 1.046). A cut-off at 2.19% yielded a sensitivity of85.7% and a specificity of 81.8% for PTSD. Positive and negativepredictive values were respectively 75% and 90%.

4. Discussion

This study is the first to investigate 24-h HRV analysis in RTAvictims as a predictor of PTSD development. Our findings seem tosupport our hypothesis as we showed that the depressed temporalparameters of HRV during the acute post traumatic phase not onlypredicted the PTSD development but also the severity of the disease

) evaluated at 2 and 6 months ((PTSDþ, patients with PTSD syndrome; PTSD�, patientsal intervals; SDNN, standard deviation of normal-to-normal intervals; ms, milliseconds;ex, percentage of RR differences).

Table 3Spearman correlation coefficients between the CAPS total scores and HRVparameters.

Parameter 2 months 6 months

Correlation coefficient p Correlation coefficient p

pNN50 �0.49 0.02 �0.47 0.05rMSSD �0.51 <0.02 �0.47 0.05Variability Index �0.47 0.03 �0.55 <0.02SDANN �0.38 0.09 �0.22 0.38SDNN �0.50 0.02 �0.39 0.11

PTSD, post traumatic stress disorder syndrome; PTSDþ, patients with PTSDsyndrome; PTSD�, patients free of PTSD syndrome; pNN50, proportion of successivenormal-to-normal interval differences >50; rMSSD, root-mean-square successivedifferences of normal-to-normal intervals; Variability Index, percentage of RRdifferences; SDANN, standard deviation of average 5-min of normal-to-normalintervals; SDNN, standard deviation of normal-to-normal intervals.

A. Shaikh al arab et al. / Journal of Psychiatric Research 46 (2012) 790e796794

in the short and long term. We focused on temporal domainanalysis of HRVwhich is the ideal method for long-term recordings.

Even though a substantial part of the literature reported thatinitially elevated HR following trauma was associated with laterPTSD, its use as an early predictor of PTSD requires sufficientsensitivity and specificity, which is not demonstrated in anyreported studies (Bryant, 2006). We questioned the utility ofaverage basal HR over 24 h as a predictor of subsequent PTSD in RTAvictims. In line with previous studies, we found a relevant associ-ation between basal mean HR measure and acute PTSD develop-ment (Coronas et al., 2011; Kraemer et al., 2008; Veazey et al.,2004). However, we found no significant effect in mean HRmeasure on chronic PTSD. The non-significant effects in the HRmeasure at six months could be due to the small sample size whichmight have led to low statistical power. However, we would notsuspect that effect as there is not even a trend in HRmeasures at sixmonths and the difference does not seem clinically relevant in ourstudy sample. We assessed the HR by taking into account the meanHR over 24 h on the second and third days of hospital admission. Inother studies, the HR was obtained from participants’ medicalcharts, by single or multiple HR assessment. Moreover, we did notfind a significant correlation between mean HR and PTSD severity,either at 2 or at 6 months. In concordance with the mixed results ofcurrent studies in this field, initial HR measured in a clinical settingis a weak and non independent predictor of PTSD following acci-dental injuries. Additionally, a meta-analysis of neuroimagingstudies identified two neurophysiological responsitivity patternsprovoked by trauma cues: an arousal subtype with approximately70% of PTSD individuals having elevated HR and a dissociative onewith 30% of those with PTSD having no HR increase (Lanius et al.,2006).

We found that psycho-traumatized victims who developedPTSD had significantly lower temporal HRV parameters than thosewithout any psychopathologic disorders in the short and long term.Our results are in favour of the diagnostic utility of HRV, which canbe considered as a more precise physiological evaluation of HRfunctioning than mean HR, to identify people at risk of PTSDfollowing RTA. The published data suggest that lower HRV can bean early reliable marker for PTSD development after RTA. Indeed,our findings seem to be related to fear-conditioningmodels of PTSDbecause diminished HRV may reflect an unconditioned response inthe acute phase of trauma exposure, as a result of dysregulation inparasympathetic activity rather than elevated sympathetic activa-tion. It is well known that basal HR is mainly under para-sympathetic control, thus strong parasympathetic modulationmaintains cardiovascular control in terms of beat-to-beat adjust-ment according to environmental demands, whereas reducedparasympathetic modulation hinders the individual’s ability to

adapt. The polyvagal theory of Porges is in line with our assump-tion. It states that baseline level of cardiac vagal tone and reactivityabilities are associated with behavioural measures of reactivity, theexpression of emotion and self-regulation skills. Therefore, thecardiac vagal tone is proposed as an index of emotion regulation.Higher amplitude HRV promotes autonomic homeostasis and refersto the ability of emotional self-regulation (Porges, 1995).

The literature often demonstrated an alteration in HRVmeasurement in PTSD, both in terms of basal level and changes inresponse to stress-related cues. Most of the previous studiesassessed short periods using frequency domain analysis of HRV,which requires highly standardized conditions and is mostlyapplied in basic psycho-physiological research (Kleiger et al., 2005).Obviously, there is a lack of studies based on long-term recordingsfor HRV assessment in patients with PTSD. While time domain HRVanalysis is a direct, simple and practical method of assessingautonomic function, frequency domain indices are log transformedbecause of their highly skewed distribution which in turn leads toweakness of statistical power (Kuss et al., 2008). A time domainmeasure such as SDNN has higher precision; it also reflects theoscillating influences of both sympathetic and parasympatheticsystems on the cardiac system. Moreover, RMSSD and pNN50 arestrongly correlated with the high frequency (HF) signal and aremore specific indicators of parasympathetic activity, providinganother measure of vagal tone (Kleiger et al., 1991). The prognosticinformation incorporates both alterations in autonomic tone andlonger term components which are best assessed using Holter ECGrecordings. HRV measured in nominal 24-h recordings is a strongrisk predictor (Task Force, 1996; Lahiri et al., 2008).

Our findings are consistent with a previous study documentinga significant link between autonomic disturbances during sleep,evaluated by frequency domain HRV analysis, and PTSD outcome at2 months in recently psycho-traumatised subjects (Mellman et al.,2004). This study had nearly the same sample size and PTSDprevalence as our study (10 PTSD�, 6 PTSDþ, 3 PTSD sub-threshold). However, the variable time delay of night recordingsfor HRV analysis was between 5 and 30 days after the trauma(18.5 � 8.8 days). Regarding the present study, we considered thesame time delay for all Holter recordings used for HRV analyseswhich were carried out on the second day of the accident. Thiscould account for the interest of the results and give them greaterreliability.

In addition, the depression has also been associated withreduced HRV (Kemp et al., 2010). Depression (as a comorbid diag-nostic status) or a depressed mood status was initially assessed inour study and then reassessed 2 and 6 months later by using theM.I.N.I. We found that it was not probable that depression influ-enced HRV measures in PTSD, as we had a low incidence ofdepression in the study sample.

Finally, our study indicated that a decrease in HRV was associ-ated with an increase in PTSD severity, as we found a negativecorrelation at both evaluations (2 and 6 months) between para-sympathetically mediated HRV indices and the severity ofPTSD symptoms assessed with the CAPS. This finding suggestsa pathological link between autonomic dysfunction and diseaseprogression.

4.1. Study limitations

The findings of this study are considered preliminary, owing tothe small number of participants and potential study limits. Giventhat age and HR were the major determinants of HRV according toFramingham Heart Study (Tsuji et al., 1996), subjects who partici-pated in our study were aged 18 years or over with no upper agelimit. However, there was no significant difference of age between

A. Shaikh al arab et al. / Journal of Psychiatric Research 46 (2012) 790e796 795

subgroups. We have to note also that assessing CAPS with a tele-phonic survey is not as good as face-to-face interviews. Any way,the CAPS was administered by well trained clinicians in PTSDassessment, and we did have PTSD prevalence comparable to thatobtained by face-to-face interviews in a recent similar study(Coronas et al., 2011). Other limitations include continued use ofnarcotic analgesics (morphine) during the ECG recording whichwould influence the ANS through possible noradrenergic attenua-tion with a trend towards vagal activation which could have playedagainst our results. Morphine would limit fear conditioning,resulting in reducing the incidence of subsequent PTSD. However,a recent study reported that acute administration of morphine inthe aftermath of traumatic injury had some protective effectagainst PTSD severity but not PTSD diagnosis (Bryant et al., 2009),this protective effect being explained by another possible mecha-nism through pain relief. Most participants in this study hadsurgical intervention under general anaesthesia following theaccidents that could influence HRV measurements. Otherwise,surgical intervention was equivalently distributed between groupsof PTSD (with/without), as all participants were RTA survivors withphysical injuries and thus, it would be unlikely to have any signif-icant influence on the results.

5. Conclusion

Our results provide a preliminary support for the utility oftemporal parameters of HRV measurement in the aftermath oftrauma exposure to identify individuals at a high risk of subsequentPTSD development in the short and long term. We suggest thatearly disturbances of ANS regulation, in particular the para-sympathetic activity, contribute to the pathogenesis of PTSD.

Role of funding source

This study was sponsored partly by Traumapsy association, andhospital clinical research grants (PHRC). They had no further role instudy design; in the collection, analysis and interpretation of data;in the writing of the report; and in the decision to submit the paperfor publication.

Contributors

AS managed the literature searches, analyses and wrote the firstdraft of the manuscript. LG guided the results analysis and vali-dated HRV analysis. François Ducrocq, Sylvie Molenda and Sté-phane Duhem conducted patient’s recruitment and psychiatricstatus assessment. Julia Salleron undertook the statistical analysisvalidation. Isabelle Chaudieu, Yvonne Fortecoeffe and Dina Bertassisted with proof-reading of the manuscript. Christian Libersasupervised and participated in study design. Guillaume Vaivadesigned the study, wrote the protocol and directed the work. Allauthors contributed to and have approved the final manuscript.

Conflicts of interest

The authors have no potential conflicts of interest.

Acknowledgement

The authors would like to thank Philippe Laffargue (Trauma-tology Department of CHR Lille), Michel Berger (TraumatologyDepartment of Douai Hospital), Anne-Laure Demarty (ClinicalInvestigation Center), Christelle Rosenstrauch (Psychiatry Depart-ment of CHR Lille), Jean-Jacques Dujardin (Cardiology Department

of Douai Hospital) and Yvonne Fortecoeffe for their contributions tothis study.

References

Aziz MA, Kenford S. Comparability of telephone and face-to-face interviews inassessing patients with posttraumatic stress disorder. Journal of PsychiatricPractice 2004;10:307e13.

Blake DD, Weathers FW, Nagy LM, Kaloupek DG, Gusman FD, Charney DS, et al. Thedevelopment of a clinician-administered PTSD scale. Journal of Traumatic Stress1995;8:75e90.

Blanchard EB, Hickling EJ. After the crash. Washington, DC: American PsychologicalAssociation; 2004.

Blanchard EB, Hickling EJ, Galovski T, Veazey C. Emergency room vital signs andPTSD in a treatment seeking sample of motor vehicle accident survivors. Journalof Traumatic Stress 2002;15:199e204.

Blechert J, Michael T, Grossman P, Lajtman M, Wilhelm FH. Autonomic and respi-ratory characteristics of posttraumatic stress disorder and panic disorder.Psychosomatic Medicine; 2007:935e43.

Bonnet MH, Arand DL. Heart rate variability in insomniacs and matched normalsleepers. Psychosomatic Medicine 1998;60:610e5.

Brunet A, Weiss DS, Metzler TJ, Best SR, Neylan TC, Rogers C, et al. The PeritraumaticDistress Inventory: a proposed measure of PTSD criterion A2. American Journalof Psychiatry 2001;158:1480e5.

Bryant RA, Creamer M, O’Donnell M, Silove D, McFarlane AC. A multisite study ofinitial respiration rate and heart rate as predictors of posttraumatic stressdisorder. Journal of Clinical Psychiatry 2008;69:1694e701.

Bryant RA, Creamer M, O’Donnell M, Silove D, McFarlane AC. A study of theprotective function of acute morphine administration on subsequent post-traumatic stress disorder. Biological Psychiatry 2009;65:438e40.

Bryant RA. Early predictors of posttraumatic stress disorder. Biological Psychiatry2003;53:789e95.

Bryant RA. Longitudinal psychophysiological studies of heart rate: mediating effectsand implications for treatment. Annals of New York Academy of Sciences 2006;1071:19e26 [Review].

Buckley B, Nugent N, Sledjeski E, Raimonde AJ, Spoonster E, Bogart LM, et al.Evaluation of initial posttrauma cardiovascular levels in association with acutePTSD symptoms following a serious motor vehicle accident. Journal of Trau-matic Stress 2004;17:317e24.

Buckley TC, Kaloupek DG. A meta-analytic examination of basal cardiovascularactivity in posttraumatic stress disorder. Psychosomatic Medicine 2001;63:585e94.

Cohen H, Kotler M, Matar MA, Kaplan Z, Miodownik H, Cassuto Y. Power spectralanalysis of heart rate variability in posttraumatic stress disorder patients. Bio-logical Psychiatry 1997;41:627e9.

Coronas R, Gallardo O, Moreno MJ, Suárez D, García-Parés G, Menchón JM. Heartrate measured in the acute aftermath of trauma can predict post-traumaticstress disorder: a prospective study in motor vehicle accident survivors. Euro-pean Psychiatry 2011;8:508e12.

Delahanty DL, Raimonde AJ, Spoonster E, Cullado M. Injury severity, prior traumahistory, urinary cortisol levels, and acute PTSD in motor vehicle accidentvictims. Journal of Anxiety Disorders 2003;17:149e64.

Hopper JW, Spinazzola J, Simpson WB, Van der Kolk A. Preliminary evidence ofparasympathetic influence on basal heart rate in posttraumatic stress disorder.Journal of Psychosomatic Research; 2006:83e90.

Kassam-Adams N, Garcia-España JF, Fein JA, Winston FK. Heart rate and post-traumatic stress in injured children. Archives of General Psychiatry 2005;62:335e40.

Keary TA, Hughes JW, Palmieri PA. Women with posttraumatic stress disorder havelarger decreases in heart rate variability during stress tasks. InternationalJournal of Psychophysiology; 2009:257e64.

Kemp AH, Quintana DS, Gray MA, Felmingham KL, Brown K, Gatt J. Impact ofdepression and antidepressant treatment on heart rate variability: a review andmeta-analysis. Biological Psychiatry 2010;67:1067e74.

Kleiger RE, Bigger JT, Bosner MS, Chung MK, Cook JR, Rolnitzky LM, et al. Stabilityover time of variables measuring heart rate variability in normal subjects.American Journal of Cardiology 1991;68:626e30.

Kleiger RE, Stein PK, Bigger JT. Heart rate variability: measurement and clinicalutility. Annals of Noninvasive Electrocardiology 2005;10:88e101 [Review].

Kraemer B, Moergeli H, Roth H, Hepp U, Schnyder U. Contribution of initial heartrate to the prediction of posttraumatic stress symptom level in accident victims.Journal of Psychiatric Research 2008;42:158e62.

Kubzansky LD, Koenen KC, Spiro A, Vokonas PS, Sparrow D. Prospective study ofposttraumatic stress disorder symptoms and coronary heart disease in theNormative Aging Study. Archives of General Psychiatry 2007;64:109e16.

Kuhn E, Blanchard EB, Fuse T, Hickling EJ, Broderick J. Heart rate of motor vehicleaccident survivors in the emergency department, peritraumatic psychologicalreactions, ASD, and PTSD severity: a 6-month prospective study. Journal ofTraumatic Stress 2006;19:735e40.

Kuss O, Schumann B, Kluttig A, Greiser KH, Haerting J. Time domain parameters canbe estimated with less statistical error than frequency domain parameters inthe analysis of heart rate variability. Journal of Electrocardiology 2008;41:287e91.

A. Shaikh al arab et al. / Journal of Psychiatric Research 46 (2012) 790e796796

Lahiri MK, Kannankeril PJ, Goldberger JJ. Assessement of autonomic function incardiovascular disease: physiological basis and pronostic implications. Journalof American College of Cardiology 2008;51:1725e33 [Review].

Lanius RA, Bluhm R, Lanius U, Pain C. A review of neuroimaging studies in PTSD:Heterogeneity of response to symptom provocation. Journal of PsychiatricResearch 2006;40:709e29.

Lecrubier Y, Sheehan D, Weiller E, Amorin P, Bonora I, Sheehan K. The Mini Inter-national Neuropsychiatric Interview (M.I.N.I.), a short diagnostic interview:reliability and validity according to the CIDI. European Psychiatry 1997;12:232e41.

Mellman TA, Knorr BR, Pigeon WR, Leiter JC, Akay M. Heart rate variability duringsleep and the early development of posttraumatic stress disorder. BiologicalPsychiatry 2004;55:953e6.

Miu AC, Heilman RM, Miclea M. Reduced heart rate variability and vagal tone inanxiety: trait versus state, and the effects of autogenic training. AutonomicNeuroscience 2009;145:99e103.

Porges SW. Orienting in a defensive world: mammalian modifications of ourevolutionary heritage. A Polyvagal Theory. Psychophysiology 1995;32:301e18[Review].

Sack M, Hopper JW, Lamprecht F. Low respiratory sinus arrhythmia and prolongedpsychophysiological arousal in posttraumatic stress disorder: heart ratedynamics and individual differences in arousal regulation. Biological Psychiatry2004;1(55):284e90.

Sahar T, Shalev AY, Porges SW. Vagal modulation of responses to mental challengein posttraumatic stress disorder. Biological Psychiatry; 2001:637e43.

Smith T. Trauma index revisited; a better triage tool. Critical Care Medicine 1990;18:174e80.

Task Force of the European Society of Cardiology and the North American Society ofPacing and Electrophysiology. Heart rate variabilitydstandards ofmeasurement,physiological interpretation, and clinical use. Circulation 1996;93:1043e65.

Tsuji H, Venditti Jr FJ, Manders ES, Evans JC, Larson MG, Feldman CL, et al. Deter-minants of heart rate variability. Journal of American College of Cardiology1996;28:1539e46.

Vaiva G, Walter M, Al Arab AS, Courtet P, Bellivier F, Demarty AL, et al. ALGOS: thedevelopment of a randomized controlled trial testing a casemanagement algorithmdesigned to reduce suicide risk amongsuicideattempters. BMCPsychiatry2011;11:1.

Vaiva G, Thomas P, Ducrocq F, Fontaine M, Boss V, Devos P, et al. Low posttraumaGABA plasma levels as a predictive factor in the development of acute post-traumatic stress disorder. Biological Psychiatry 2004;55:250e4.

Veazey CH, Blanchard EB, Hickling EJ, Buckley TC. Physiological responsiveness ofmotor vehicle accident survivors with chronic posttraumatic stress disorder.Applied Psychophysiology and Biofeedback 2004;29:51e62.

Zatzick DF, Russo J, Pitman RK, Rivara F, Jurkovich G, Roy-Byrne P. Reevaluating theassociation between emergency department heart rate and the development ofposttraumatic stress disorder: a public health approach. Biological Psychiatry2005;1(57):91e5.

Zucker TL, Samuelson KW, Muench F, Greenberg MA, Gevirtz RN. The effects ofrespiratory sinus arrhythmia biofeedback on heart rate variability and post-traumatic stress disorder symptoms: a pilot study. Applied Psychophysiologyand Biofeedback 2009;34:135e43.