Association between PTSD and Impedance Cardiogram-based Contractility Metrics During Trauma Recall: A Controlled Twin Study

Abstract

Post-traumatic stress disorder (PTSD) is an independent risk factor for incident heart failure, but the underlying cardiac mechanisms remained elusive. Impedance cardiography (ICG), especially when measured during stress, can help understand the underlying psychophysiological pathways linking PTSD with heart failure. We investigated the association between PTSD and ICG-based contractility metrics (pre-ejection period (PEP) and Heather index (HI)) using a controlled twin study design with a laboratory-based traumatic reminder stressor. PTSD status was assessed using structured clinical interviews. We acquired synchronized electrocardiograms and ICG data while playing personalized-trauma scripts. Using linear mixed-effects models, we examined twins as individuals and within PTSD-discordant pairs. We studied 137 male veterans (48 pairs, 41 unpaired singles) from Vietnam War Era with a mean (standard deviation) age of 68.5(2.5) years. HI during trauma stress was lower in the PTSD vs. non-PTSD individuals (7.2 vs. 9.3 [ohm/s²], p =0.003). PEP reactivity (trauma minus neutral) was also more negative in PTSD vs. non-PTSD individuals (−7.4 vs. −2.0 [ms], p= 0.009). The HI and PEP associations with PTSD persisted for adjusted models during trauma and reactivity, respectively. For within-pair analysis of eight PTSD-discordant twin pairs (out of 48 pairs), PTSD was associated with lower HI in neutral, trauma, and reactivity, whereas no association was found between PTSD and PEP. PTSD was associated with reduced HI and PEP, especially with trauma recall stress. This combination of increased sympathetic activation and decreased cardiac contractility combined may be concerning for increased heart failure risk after recurrent trauma re-experiencing in PTSD.

1. Introduction

Post-traumatic stress disorder (PTSD) is a chronic psychiatric illness that arises from exposure to a traumatic event and manifests itself in intrusive memories of the trauma, hypervigilance, and hyperarousal (Yehuda et al., 2015). The lifetime prevalence of PTSD in the U.S. is approximately 7% (Kessler et al., 2005) with prevalence ranging from 20% to 30% among U.S. veterans (Trivedi et al., 2015; Zhu et al., 2021). Previous studies have demonstrated that PTSD is an independent risk factor for developing heart failure, however, the underlying mechanism is not clear. Roy et al., (2015) conducted the first large-scale longitudinal study of 8,248 US veterans which showed a positive association between PTSD and incident heart failure. Taylor-clift et al., (2016) also found an association between PTSD and earlier onset of heart failure among 251 low-income heart failure patients. Song et al., (2019) conducted a population-based, sibling-controlled study in which the stress-related disorders including PTSD were associated with a 6.94 hazard ratio of heart failure within the first year of its diagnosis.

The autonomic nervous system (ANS) controls and regulates the cardiovascular system (Jänig, 2008). ANS further consists of the parasympathetic nervous system (PNS) and sympathetic nervous system (SNS) which are responsible for “rest and digest” and “fight or flight” responses, respectively. A balanced interaction between the PNS and SNS is required for the normal functioning of ANS (Johnson, 2018). However, ANS functioning is severely affected by PTSD (Orr &Roth, 2000). In PTSD participants, higher levels of dopamine and norepinephrine concentration are observed, which may be a reason for the change in ANS activity (Yehuda et al., 1992). A lower PNS activity was observed in PTSD participants via analysis of heart rate variability (HRV) metrics of root mean square of successive differences (RMSSD), and high frequency (HF) (Schneider & Schwerdtfeger, 2020). Heart rate (HR) was found to be higher in the resting state for PTSD participants as compared to the non-PTSD participants (Pole, 2007). PTSD participants have also shown either blunted or increased cardiovascular response to stressful tasks (Cohen et al., 2000; Dennis et al., 2016; Jovanovic et al., 2009).

SNS activation has been described in PTSD during trauma reminders (Pole, 2007). Psychophysiological responsivity in individuals with PTSD, as they recall their personal traumatic events in a laboratory setting (“trauma recall”), can help in understanding the linkage between PTSD and heart diseases (Schmahl et al., 2004), capturing pathophysiologic processes (Bauer et al., 2013), estimating PTSD symptom severity (Castro-Chapman et al., 2018), and evaluating treatment progress and outcome (Wangelin & Tuerk, 2015; Katz et al., 2020; Raskind et al., 2016). Previous studies examining HR and blood pressure (BP) changes during trauma recall are relatively nonspecific and unfortunately offer limited insight on specific diseases such as heart failure (Pole, 2007; Bauer et al., 2013; Castro-Chapman et al., 2018; Wangelin & Tuerk, 2015; Katz et al., 2020). HR is influenced by both sympathetic and parasympathetic influences, whereas BP (systolic (SBP) and diastolic (DBP)) is affected by both cardiac contractility and total peripheral resistance (Bootsma et al., 2003; Czarnek et al, 2021). Also, the HRV metrics do not provide an independent SNS activity analysis. The highly correlated metrics of RMSSD and HF-HRV represent PNS activity (Thayer & Lane, 2007). The standard deviation of the normalized NN-intervals (SDNN) is influenced by both PNS and SNS activity (Shaffer & Ginsberg, 2017). The association of LF-HRV with SNS is controversial and it is thought to reflect PNS activity, at least in part, as well as baroreflex sensitivity (Goldstein et al., 2011).

Through impedance cardiogram (ICG), the first-order time derivative of the thoracic impedance signal Z(t) (i.e., dZ/dt), one can assess cardiac mechanical function and help to understand more specific mechanisms of stress that relate to heart failure (Sherwood et al., 1990). The ICG specifically yields contractility metrics that are more sensitive to changes in the sympathetic impact on the heart than HR or BP (Berntson et al., 2007). SNS activation has been described in PTSD during trauma reminders (Pole, 2007) and, in turn, may affect ICG-based contractility metrics (pre-ejection period (PEP) and Heather index (HI)), which are valid indices of sympathetic beta-adrenergic activity (Newlin & Levenson, 1979; Thayer et al., 2010).

PEP is inversely related to myocardial contractility and analyzed in various stress studies (Kelsey et al., 2007; Krohová et al., 2017; Peters et al., 2018; Cohen et al, 2020), although examinations specifically in PTSD are rare. HI is an index of left ventricular contractility function. Initially, it was computed as (dZ/dt)_max/QC, where (dZ/dt)_max was the maximum amplitude of ICG (C point on ICG) and the QC interval was the time interval between Q point on ECG and C point on ICG signal (Heather, 1969). The HI approximates the blood acceleration in the aorta during systole, and is measured in units of ohm/sec² (Baker et al., 1983). More recently, HI has been measured as (dZ/dt)_max/RC (Peng et al. 2004), which has the advantage of the ease in the detection of R peak on the ECG signal.

Laboratory experiments with animals have demonstrated that HI is correlated with the acceleration of blood through the aorta and represents the pumping ability of the heart. Reductions in HI values are 4 to 5 times more sensitive in detecting contractility than other measures like PEP and left ventricular ejection time (Baker et al., 1983; Peng et al., 2004). Low HI values were observed in patients with clinical evidence of heart failure as well (Hubbard et al., 1986). By precisely estimating early systolic excursion velocities of the LV, it may help to understand subclinical abnormalities in left ventricular function. As further evidence of its predictive abilities in HF, it has also been associated with ventricular wall tension, the maximum rate of pressure-rise in the ventricle (Thompson et al., 1981), and 2D fractional shortening (Jurn & Heethaar, 1999). Therefore, evaluation of HI and PEP in PTSD can provide important insights into its cardiac autonomic and cardio-mechanical effects.

In this study, we examined the relationship between PTSD and ICG-based contractility metrics within twin pairs that control for numerous shared genetic, familial, and environmental factors (McGue et al., 2010; Carlin et al., 2005). Using a laboratory trauma recall stress, we experimentally invoked acute PTSD symptoms and examined their effects on left ventricular autonomic and mechanical function by assessing examining PEP and HI. Although the study focus was the analysis of the ICG-based metrics (PEP, HI), we also analyzed HR, RMSSD, BP (SBP, DBP), and total peripheral resistance (TPR) to ensure a thorough physiological assessment. For an in-depth analysis of the HRV metrics (deceleration capacity (DC), high frequency (HF) power, and logarithmic low frequency (log-LF) power) under laboratory-trauma recall stress, the interested readers are directed to a study by Perez, et. al, (2022), which revealed that PTSD status and acute PTSD symptom severity were significantly associated with the lower DC and log-LF during the traumatic recall.

In this study, we hypothesized that PTSD is associated with lower PEP (indicating higher SNS activity) and lower HI (indicating reduced blood acceleration through the aorta) during the neutral phase, and that these associations strengthen during trauma recall when pathological neurocardiac circuits may be more activated and underlying cardiac pathologies may be revealed.

2. Method

2.1. Study Cohort

This study was performed as an ancillary project on the individuals from the Emory Twin Follow-up Study (ETFS) conducted from 2016 to 2019 (Vaccarino et al, 2022). The study included 279 male veterans (124 twin pairs, 31 unpaired singles) from Vietnam War Era. We oversampled for pairs of twins who were discordant for depression and PTSD during a prior examination. A clinical diagnosis of PTSD was obtained using the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorder, 4^th Edition (SCID) (First et al., 2004). We introduced ICG as part of an ancillary study that was funded after the primary study (focused on cardiac imaging) was underway. As such, of the 279 (n = 67 with PTSD) participants who enrolled in ETSF, we obtained ICG data from 167 of them. From this sample, we could not analyze data from 30 participants because of file corruption (n=3), errors in signal acquisition (n=9), missing event markers (n=11), and poor signal quality (n=7). Our final sample for this analysis included 137 male veterans (48 twin pairs, 41 unpaired singles) with a mean (SD) age of 68.5 (2.5) years. Written informed consent was taken from all participants and the study was approved by the Emory Institutional review board number IRB00081004.

2.2. Study Protocol

The study measured ICG-based contractility in response to neutral and trauma audio sessions. There was no formal resting baseline, and all participants listened to the neutral scripts first, followed by their recorded trauma. In the neutral session of 4 to 7 minutes, the participants listened to two neutral audio scripts. Participants hand-wrote one-page descriptions of the two most traumatic events in their memory the day before the trauma recall stress challenge. Study staff members who were not acquainted with the participants recorded one-minute audio clips that read all parts of the trauma in the veteran’s own words. The next day, these audio recordings were played back to the participants with a five-minute break between sessions. The participants then completed a 14-item PTSD symptoms scale (PTSDSS) for the assessment of PTSD symptom severity after the trauma recall stress challenge (Southwick et al., 1993). A lab technician logged the time stamps for the start/end of audio scripts and sessions in real-time.

2.3. Measurement of ICG-based Contractility Metrics

The participants were seated in a comfortable chair, and synchronized ECG and ICG signals were recorded at a sampling frequency of 1000 Hz by using ECG and ICG equipment from BIOPAC (BIOPAC Systems Inc. Goleta, CA). ECG signals were acquired via three disposable 3 M red dot electrodes (two attached to the collar bone, and the ground electrode attached to the hip). Four disposable EL507 spot electrodes were used to record the ICG signal. An alternating current of 400 microampere at 50 kHz was injected into the outer current electrodes located at the side of the neck and the left lateral side of the thorax below the xiphisternal junction. To pick up the voltage, an inner electrode was placed below the outer current electrode at the neck side, and the other inner electrode was placed above the outer current electrode at the left lateral side of the thorax.

AcqKnowledge 5.0 software was used for data storage, display, and data export functions. MATLAB R2017b (MathWorks, Inc., Natick, MA) was used for pre-processing the signals and applying our open-source automated ICG noise removal algorithm (Three-Stage Ensemble Average Algorithm) to generate the synchronized ensemble-averaged (EA) beats over 60-seconds non-overlapping sampling interval (Sheikh, Shah, Levantsevych, et al., 2020). All annotations required for computation of PEP and HI were performed using the “Impedance Cardiogram Manual Annotation Application (ICMAA)” from our open-source ICG toolbox (Sheikh, Shah, Inan, and Clifford, 2020). Figure 1(A) presents the PEP and HI computation process.

Figure 1:

Pre-ejection period (PEP) and Heather index (HI) computation and measurement methodology. Figure 1(A) presents the signal flow for computation of PEP and HI. ECG and ICG signals were simultaneously recorded from the participant. In the pre-processing stage, the signals were band-pass filtered with lower and upper cutoff frequencies of 0.5 and 40 Hz respectively. Synchronized averaged beats were generated using the noise removal algorithm. Impedance cardiogram manual annotation application was used to annotate the critical points and compute PEP and HI. Figure 1(B) presents the measurement of PEP and HI using the synchronized ICG and ECG beats. PEP was computed as time interval maximum depolarization left ventricular (R peak on ECG) and opening of aortic valve (B point on ICG). HI was measured as ratio of maximum amplitude of ICG (dZ/dt_max) and time interval RC. Image (c) Emory University, CC-BY-SA.

PEP, measured in milliseconds (ms), was computed as the time interval between maximum depolarization of left ventricular (R peak on EA ECG beat) and opening of the aortic valve (B point on EA ICG beat) (Seery et al., 2016). HI, measured as (dZ/dt)_max/RC in ohm/sec², was computed as ratio of maximum amplitude of ICG representing maximum blood velocity in aorta (amplitude of C point on EA ICG beat), and time interval between R peak and C point on EA ECG and ICG beats respectively (Peng et al., 2004). Figure 1(B) depicts the measurements of PEP and HI via synchronized ECG and ICG waveforms.

R peaks on EA ECG beats were automatically detected via the PhysioNet Cardiovascular Signal Toolbox using the “jqrs” algorithm (Vest et al., 2018). Three expert physicians independently performed the B-point manual annotation on the EA ICG beats by following the ICG guidelines (Sherwood et al., 1993; Árbol et al., 2017). The C point was automatically detected as the highest point on the EA ICG beat in the one-third duration of the RR interval from the R peak of the synchronized EA ECG beat. All annotations were reviewed collectively by the experts and edited to ensure accurate scoring using ICMAA.

The contractility metrics were computed from each EA ECG/ICG beat in the neutral and trauma sessions. Later, the metrics were averaged over the neutral and trauma sessions separately for each participant. The reactivity metrics (trauma minus neutral) for each participant were computed by subtracting the averaged neutral session metrics from the averaged trauma session metrics.

2.4. Measurement of non-ICG-based Physiological Metrics

To ensure a thorough physiological assessment, we also measured non-ICG-based physiological metrics including HR, RMSSD, BP (SBP, DBP) and TPR for each session separately. The HR measurements were derived from the ECG signal over 60-seconds non-overlapping sampling interval. RMSSD was computed over 60-seconds window length using an increment of 10 seconds via PhysioNet Cardiovascular Signal Toolbox (Vest et al., 2018). The BP measurements were taken at the end of each neutral and trauma audio script using an automatic oscillometric device (Datascope Accutorr). TPR was computed as the ratio of mean arterial pressure (MAP) and cardiac output (CO) (Hill et al., 2013). MAP was measured as DBP + (1/3)*(SBP-DBP). CO was measured as [(SBP -DBP)/(SBP + DBP)]*HR (Liljestrand and Zander, 1928). All the measurements were averaged for each session separately.

2.5. Other Measurements

A thorough assessment of the sociodemographic factors, lifestyle factors, cardiovascular risk factors, psychiatric diagnosis, and current medications was performed as previously described (Vaccarino et al., 2011). We used the Baecke Questionnaire of Habitual Physical Activity to measure physical activity (Richardson et al., 1995). We defined diabetes mellitus as having measured fasting glucose >126 mg/dL or being treated with antidiabetic medications. We defined the history of hypertension as SBP >140 mm Hg or DBP >90 mm Hg, or self-reported use of antihypertensive medications. A history of coronary artery disease that might have occurred from the time of the initial screen was defined as a diagnosis of myocardial infarction or coronary revascularization procedures. Subclinical coronary heart disease (CHD) was assessed using Positron Emission Tomography (PET) myocardial perfusion imaging to identify any unreported evidence of heart disease, which was defined as at least a 5% defect on stress or rest images to indicate evidence of previous infarction or ischemia. The administration of SCID also provided clinical diagnoses of lifetime history of substance abuse disorder and of major depressive disorder. We used military records to determine service in Southeast Asia. The variable, “service in Southeast Asia”, indicates the deployment of the participants in Vietnam, and considers the additional exposures that could be confounding factors for developing PTSD.

2.6. Statistical Analysis

We conducted descriptive analyses by summarizing participants’ characteristics related to sociodemographic factors, lifestyle factors, cardiovascular risk factors, psychiatric diagnosis, current medications, and PTSDSS.

We compared the ICG-based contractility indices and non-ICG-based physiological metrics between PTSD and non-PTSD participants by considering them as individual twins in the neutral, trauma, and reactivity analysis. We used mixed-effects models and accounted for twin-pair as a random effect. We examined the impact of apriori selected factors on the association analysis by constructing a succession of four models. In all models, PTSD status was primary exposure, and the outcomes were the ICG-based contractility metrics (PEP or HI) or non-ICG-based physiological metrics (HR, RMSSD, SBP, DBP, or TPR). The four models are:

• Model 1: This was the initial unadjusted model where PTSD status was the only exposure.

• Model 2: In this model, we further adjusted for sociodemographic and lifestyle factors (married, employment, service in Southeast Asia, current smoking status, physical activity).

• Model 3: In this model, we further adjusted for cardiovascular risk factors (body mass index, diabetes, hypertension, history of CHD, cardiac PET-based perfusion defects) and medications (current use of aspirin, statin, beta-blocker, and angiotensin converting enzyme medications).

• Model 4: In this model, we further adjusted for psychiatric diagnosis (lifetime history of major depressive disorder, substance abuse, and current use of antidepressants.).

In order to examine the relationship of all metrics with acute PTSD symptoms, we also analyzed the associations between the PTSDSS and all metrics for neutral, trauma, and reactivity separately. We used mixed-effects models and accounted for twin-pair as a random effect. We examined the impact of apriori selected factors on the association analysis by constructing a succession of four models like that of within-pair difference of PTSD status. In all models, PTSDSS was primary exposure, and the outcomes were the ICG-based contractility metrics (PEP or HI) and non-ICG-based physiological metrics (HR, RMSSD, SBP, DBP, or TPR).

We also compared all indices between the PTSD discordant twin pairs. We used the linear mixed-effects models to analyze the associations between the within-pair difference of PTSD status and the within-pair difference of all metrics by accounting for twin-pair as a random effect. The linear mixed-effects model has an inherent advantage of not requiring a balance of the data (Cnaan & Slasor, 1997; Littell, 2002). The analysis was performed for neutral, trauma, and reactivity separately. The within-pair difference was computed as the difference of each individual from the twin-pair mean values (Carlin et al., 2005). In a twin study, the within-pair differences inherently control for potential confounding by demographic and shared familial influences, as well as environmental factors during the clinic visit as twins were examined together (McGue et al., 2010).

In addition, we also examined the relationship of all metrics with acute PTSD symptoms by analyzing the associations between the within-pair difference of PTSDSS and within-pair difference of ICG-based contractility metrics for neutral, trauma, and reactivity separately. We used mixed-effects models and accounted for twin-pair as a random effect, where PTSDSS was the exposure, and the outcomes were the ICG-based contractility metrics (PEP or HI), and non-ICG-based physiological metrics (HR, RMSSD, SBP, DBP, or TPR).

Among the 120 non-PTSD participants, 15 participants had remitted PTSD. We also examined the relationship between three-level PTSD status (current PTSD (n = 17), past PTSD (n =15), and no PTSD (n =105)) and all indices (PEP, HI, HR, RMSSD, SBP, DBP, and TPR) in the same models for the participants as individuals as well as twin pairs.

Missing data were rare (<5%); thus, we used all available data without imputation. For all mixed-effects models described above, a two-sided p-value <0.05 was used for statistical significance using maximum likelihood methods. A 95% CI was calculated around the estimate (β₁) from the model parameters (radius of the standard error times a percentage point from the t distribution). Statistical analyses were performed using SAS, version 9.4 (SAS Institute).

3. Results

3.1. Study cohort characteristics

The 137 participants included 48 pairs and 41 unpaired singles. Among 48 twin pairs, there were eight PTSD discordant pairs, one PTSD concordant pair, and 39 non-PTSD pairs. Among 41 unpaired singles, there were seven PTSD and 34 non-PTSD participants. The pairs that contained no PTSD status were likely the control pairs from the original recruitment into the first Emory Twins Study (Vaccarino et al., 2013).

Table 1 presents the comparison of characteristics between PTSD (n=17) and non-PTSD (n =120) participants analyzed in this study. Among the 120 non-PTSD participants, 15 participants were past PTSD but diagnosed as non-PTSD via SCID. The mean age and years of education were similar across both groups, but PTSD participants were less likely to be married and employed. PTSD participants were more likely to have served in Southeast Asia and to be current smokers than non-PTSD participants. The difference in the body mass index, BP (SBP, DBP), and history of hyperlipidemia were minor between both groups, but PTSD participants had a higher prevalence of diabetes. Non-PTSD participants had a higher prevalence of hypertension and coronary heart disease and were likely to take medications for the prevention of cardiovascular diseases. PTSD participants had higher evidence of CHD via myocardial perfusion imaging and were also more likely to have a diagnosis of substance abuse (alcohol and drug abuse), and major depression.

Table 1.

Sociodemographic, lifestyle, cardiovascular risk factors, psychiatric diagnosis, and respective medications by PTSD status.¹

Factors	PTSD (N = 17)	Non-PTSD (N = 120)
Sociodemographic factors
Age, years, mean (SD)	67.8 (1.6)	68.4 (2.6)
Non-white, %	12	4
Married, %	59	79
Years of education, mean (SD)	12.3 (1.4)	14.1 (2.4)
Employed full time, %	18	23
Service in Southeast Asia, %	65	36
Lifestyle factors
Cigarette Smoking
Never, %	24	35
Former, %	53	51
Current, %	24	13
Physical Activity (Baecke Score), mean (SD)	7.6 (1.1)	8.1 (1.3)
Cardiovascular risk factors and medications
BMI, mean (SD)	31.7 (5.1)	29.2 (4.2)
SBP, mm Hg, mean (SD)	139.7 (13.5)	140.9 (17.4)
DBP, mm Hg, mean (SD)	81.8 (15.3)	80 (10.7)
History of hyperlipidemia, %	65	66
History of diabetes, %	30	23
History of hypertension, %	47	54
Self-reported coronary heart disease (CHD), %	6	16
Perfusion defects >5% in size on rest or stress cardiac PET, %	42	18
Aspirin, %	24	48
Statins, %	35	61
Beta-blockers, %	18	24
ACE Inhibitors, %	6	26
Psychiatric diagnoses (lifetime) and medications
Major Depressive disorder, %	82	10
Alcohol Abuse (with or without dependence), %	47	22.5
Drug Abuse (with or without dependence), %	24	11
PTSD Severity score, mean (SD)	34.6 (12.0)	21.6 (8.2)
Antidepressants, %	65	11

¹SD = Standard deviation; mm = millimeter; Hg = Mercury; PET = Positron Emission Tomography; ACE = Angiotensin Converting Enzyme.

A comparison of characteristics between current PTSD, past PTSD, and non-PTSD participants can be found in Table A1, Appendix A. Also, Table A2, appendix A, presents the descriptive statistics for HR, RMSSD, SBP, DBP, TPR, PEP, and HI for current PTSD, past PTSD, and non-PTSD participants in the neutral condition, trauma recall, and reactivity.

3.2. Analysis of PTSD participants as individuals

We present in Table 2 the descriptive statistics for HR, RMSSD, SBP, DBP, TPR, PEP, and HI for PTSD and non-PTSD participants in the neutral condition, trauma recall, and reactivity. In general, PTSD participants had lower PEP and HI values for neutral, trauma, and reactivity as compared to the non-PTSD participants as shown in Figure 2(A) and Figure 2(B). The p-values describe the outcome of the statistical test between PTSD and non-PTSD groups based on mixed models for each type of outcome in the figure. These methods were discussed in section 2.6. Table 3 presents the multivariable analysis of the relationship between PTSD status and PEP and HI. PTSD status was significantly associated with PEP in trauma (Model 4 only) and reactivity (for all Models). PTSD status was significantly associated with HI in neutral (for all Models), and trauma (for Models 1, 2, and 3). Table 4 presents the multivariate analysis of the relationship between PTSDSS and these two ICG metrics (i.e., PEP and HI). It shows mostly non-significant results.

Table 2:

Descriptive statistics for heart rate (HR) in beats per minute (bpm), root mean square of successive difference (RMSSD) in milli-second (ms), systolic blood pressure (SBP) in milli-meter of mercury (mm Hg), diastolic blood pressure (DBP) in mm Hg, total peripheral resistance (TPR) in mmHg.min/L, pre-ejection period (PEP) in ms, Heather index (HI) in ohm/seconds² (ohm/s²) for PTSD (n =17) and non-PTSD participants (n =120) in neutral, trauma, and reactivity. The HR measurements were derived from the ECG signal over 60-seconds non-overlapping sampling interval. The BP (SBP and DBP) measurements were taken twice during the neutral and trauma sessions. Then, the BP and HR measurements were averaged for each session separately. PEP and HI were also computed over 60-seconds non-overlapping windows and averaged for each session separately. The reactivity metrics (trauma minus neutral) for each participant were computed by subtracting the averaged neutral session metrics from the averaged trauma session metrics.

Metrics	PTSD (17)			No PTSD (120)
	Mean	SD	95% CI	Mean	SD	95% CI
Neutral
HR (bpm)	72.3	11.6	[66.3, 78.3]	67.2	9.8	[65.5, 69.0]
RMSSD (ms)	35.4	33.6	[19.4, 51.4]	29.1	28.5	[23.9, 34.3]
SBP (mm Hg)	146.8	20.0	[136.5, 157.1]	145.7	24.3	[141.4, 150.0]
DBP (mm Hg)	79.6	15.8	[71.5, 87.7]	79.9	13.6	[77.5, 82.3]
TPR (mmHg.min/L)	5.1	1.6	[4.3, 5.9]	5.6	1.6	[5.3, 5.8]
PEP (ms)	68.4	21.8	[57.2, 79.6]	72.4	21.1	[68.6, 76.1]
HI (ohm/s2)	7.0	2.2	[5.8, 8.1]	9.0	2.7	[8.5, 9.5]
Trauma
HR (bpm)	75.3	12.8	[68.7, 81.9]	70.0	11.1	[68.0, 71.9]
RMSSD (ms)	34.4	29.8	[20.3, 48.6]	27.6	26.8	[22.8, 32.5]
SBP (mm Hg)	151.5	16.6	[142.9, 160.1]	153.8	25.7	[149.2, 158.4]
DBP (mm Hg)	86.9	17.5	[77.8, 95.9]	86.1	13.7	[83.6, 88.5]
TPR (mmHg.min/L)	5.8	1.8	[4.9, 6.7]	5.9	1.6	[5.6, 6.2]
PEP (ms)	60.9	18.7	[51.3, 70.6]	70.4	21.2	[66.6, 74.2]
HI (ohm/s2)	7.2	2.4	[6.0, 8.4]	9.3	3.1	[8.7, 9.8]
Reactivity
HR (bpm)	3.0	4.6	[0.6, 5.4]	2.8	4.5	[1.9, 3.6]
RMSSD (ms)	−1.0	11.5	[−6.5, 4.5]	−1.5	7.3	[−2.8, −0.1]
SBP (mm Hg)	4.7	13.6	[−2.3, 11.7]	8.1	14.0	[5.6, 10.6]
DBP (mm Hg)	7.3	9.4	[2.4, 12.1]	6.2	9.1	[4.6, 7.8]
TPR (mmHg.min/L)	0.7	0.9	[0.2, 1.1]	0.3	1.0	[0.2, 0.5]
PEP (ms)	−7.4	8.5	[−11.8, −3.1]	−2.0	5.9	[−3.1, −0.9]
HI (ohm/s2)	0.2	0.8	[−0.2, 0.6]	0.3	0.9	[0.1, 0.4]

Figure 2:

PEP and HI comparison for PTSD as individuals and within-pair analysis for neutral, trauma, and reactivity. Error bars represent 95% CI. Top panel Figures 2(A) and (B) present the comparison of PEP and HI respectively for PTSD status. The p-values depict the PTSD vs. non-PTSD comparison for PEP and HI. Bottom panel Figures 2(C) and (D) depict the comparison of within-pair (eight PTSD discordant pairs) PEP differences and within-pair HI differences, respectively, by PTSD status. The p-values represent the association for unadjusted Model-1. Image (c) Emory University, CC-BY-SA.

Table 3:

Multivariable analysis of the relationship between PTSD status and ICG metrics (Pre-ejection period (PEP) and Heather Index (HI)) for PTSD participants as individuals.

Model	Neutral		Trauma		Reactivity
	Coefficient	[95 % CI]	Coefficient	[95 % CI]	Coefficient	[95 % CI]
Outcome: PEP
Model 1a	−4.6	[−15.4, 6.1]	−9.5	[−20.2, 1.2]	−4.1	[−7.1, −1.1]
Model 2b	−2.1	[−13.2, 9.0]	−7.3	[−18.3, 3.7]	−4.7	[−7.7, −1.6]
Model 3c	−3.8	[−15.9, 8.2]	−8.0	[−20.0, 4.0]	−4.1	[−7.2, −0.84]
Model 4d	−8.3	[−22.6, 6.0]	−14.8	[−29.0, −0.53]	−6.5	[−10.3, −2.6]
Outcome: HI
Model 1a	−2.2	[−3.5, −0.82]	−2.3	[−3.9, −0.82]	−0.12	[−0.59, 0.33]
Model 2b	−2.4	[−3.8, −1.0]	−2.5	[−4.1, −0.99]	−0.09	[−0.56, 0.38]
Model 3c	−1.8	[−3.2, −0.4]	−1.7	[−3.3, −0.17]	0.04	[−0.47, 0.55]
Model 4d	−2.0	[−3.7, −0.33]	−1.9	[−3.8, 0.06]	0.17	[−0.47, 0.81]

^aModel 1 = Unadjusted base model for PTSD status

^bModel 2 = Model 1 + sociodemographic and lifestyle factors including marital status, employment, service in Southeast Asia, current smoking status, and physical activity.

^cModel 3 = Model 2 + cardiovascular risk factors including BMI, diabetics, hypertension, history of CHD, cardiac PET-based perfusion defects, and current use of aspirin, statin, beta-blocker, and angiotensin converting enzyme.

^dModel 4 = Model 3 + psychiatric diagnosis including lifetime history of major depressive disorder, substance abuse, and current use of antidepressants.

Table 4:

Multivariable analysis of the relationship between PTSDSS and ICG metrics (Pre-ejection period (PEP) and Heather Index (HI)) for PTSD participants as individuals. PTSDSS was significantly associated with PEP in reactivity for models 1 and 4 only. PTSDSS was not associated with HI.

Model	Neutral		Trauma		Reactivity
	Coefficient	[95 % CI]	Coefficient	[95 % CI]	Coefficient	[95 % CI]
Outcome: PEP
Model 1a	0.06	[−0.31, 0.43]	−0.07	[−0.44, 0.30]	−0.11	[−0.21, −0.002]
Model 2b	0.15	[−0.23, 0.52]	0.03	[−0.35, 0.40]	−0.1	[−0.2, 0.01]
Model 3c	0.04	[−0.37, 0.44]	−0.05	[−0.46, 0.35]	−0.1	[−0.2, 0.004]
Model 4d	0.002	[−0.42, 0.42]	−0.11	[−0.53, 0.31]	−0.1	[−0.3, −0.01]
Outcome: HI
Model 1a	−0.04	[−0.09, 0.01]	−0.04	[−0.09, 0.02]	0.0	[−0.01, 0.01]
Model 2b	−0.04	[−0.09, 0.01]	−0.04	[−0.09, 0.01]	0.0	[−0.02, 0.02]
Model 3c	−0.04	[−0.09, 0.01]	−0.04	[−0.09, 0.02]	0.01	[−0.01, 0.02]
Model 4d	−0.04	[−0.09, 0.01]	−0.03	[−0.09, 0.02]	0.01	[−0.01, 0.03]

^aModel 1 = Unadjusted base model for PTSDSS

^bModel 2 = Model 1 + sociodemographic and lifestyle factors including marital status, employment, service in Southeast Asia, current smoking status, and physical activity.

^cModel 3 = Model 2 + cardiovascular risk factors and respective medications including BMI, diabetics, hypertension, history of CHD, cardiac PET-based perfusion defects, and current use of aspirin, statin, beta-blocker, and angiotensin converting enzyme.

^dModel 4 = Model 3 + psychiatric diagnosis including lifetime history of major depressive disorder, substance abuse, and current use of antidepressant.

Appendix B represents the analysis of the association between PTSD status /PTSDSS and non-ICG-based metrics. PTSD status was not significantly associated with HR, RMSSD, SBP, DBP, or TPR for any of the models (Table B1, Table B2, Table B3). Table B4 presents the multivariate analysis of the relationship between PTSDSS and HR and RMSSD. PTSDSS was significantly associated with the HR during neutral and trauma stages. Table B5 presents the relationship between PTSDSS and BP metrics (SBP and DBP). PTSDSS was significantly associated with DBP during neutral and trauma recall. Table B6 presents the relationship between PTSDSS and TPR. PTSDSS was significantly associated with TPR during trauma recall for Model 2 and Model 3 only.

The analysis of the multivariate relationship between the three-level PTSD status (current PTSD, past PTSD, no-PTSD) as individuals and non-ICG-based physiological metrics revealed a significant, positive association between the three-level PTSD status and HR in neutral (Model 1, Model 3, and Model 4), and trauma (Model 3, and Model 4) only.

3.3. Within-pair PTSD analysis

Table 5 presents the descriptive statistics for eight PTSD-discordant twin pairs, which follow similar trends as the individual-level analysis. For within-pair analysis of PTSD-discordant twins, PTSD participants generally showed lower PEP and HI values than their non-PTSD brothers in neutral, trauma, and reactivity as shown in Figure 2(C) and Figure 2(D) respectively. A robust association was found between the PTSD status and HI for neutral (coefficient = −3.2; CI = [−4.5, −2.0]; p < 0.001), trauma (coefficient = −3.9; CI = [−5.3, −2.5]; p < 0.001), and reactivity conditions (coefficient = −0.7; CI = [−1.1, −0.3]; p < 0.05). No significant association was found between the PTSD status and PEP for neutral (coefficient = −7.1; CI = [−17.6, 3.3]; p = 0.2), trauma (coefficient = −8.1; CI = [−18.7, 2.5]; p =0.1) and reactivity (coefficient = −0.97; CI = [−3.4, 1.4]; p = 0.4).

Table 5:

Descriptive statistics for heart rate (HR) in beats per minute (bpm), root mean square of successive difference (RMSSD) in milli-second (ms), systolic blood pressure (SBP) in milli-meter of mercury (mm Hg), diastolic blood pressure (DBP) in mm Hg, total peripheral resistance (TPR) in mmHg.min/L, pre-ejection period (PEP) in ms, and Heather index (HI) in ohm/seconds² (ohm/s²) for PTSD-discordant pairs, comparing in neutral, trauma, and reactivity. The HR measurements were derived from the ECG signal over 60-seconds non-overlapping sampling interval. The BP (SBP and DBP) measurements were taken twice during the neutral and trauma sessions. The BP measurements were taken at the end of each neutral and trauma audio script. Then, the BP and HR measurements were averaged for each session separately. PEP and HI were also computed over 60-seconds non-overlapping windows and averaged for each session separately. The reactivity metrics (trauma minus neutral) for each participant were computed by subtracting the averaged neutral session metrics from the averaged trauma session metrics.

Metrics	Descriptive Statistics of parameters for eight PTSD discordant pairs (n =16)
	PTSD (8)			No PTSD (8)
	Mean	SD	95% CI	Mean	SD	95% CI
Neutral
HR (bpm)	70.3	5.7	[65.5, 75.1]	66.8	11.7	[58.7, 74.9]
RMSSD (ms)	28.2	26.7	[5.9, 50.5]	35.3	26.1	[18.2, 54.2]
SBP (mm Hg)	138.7	17.0	[124.5, 152.9]	147.9	24.1	[131.2, 164.6]
DBP (mm Hg)	73.0	14.8	[60.1, 85.4]	73.8	11.8	[65.6, 82.0]
TPR (mmHg.min/L)	4.5	1.3	[3.4, 5.6]	4.7	1.4	[3.8, 5.7]
PEP (ms)	69.8	20.9	[52.3, 87.4]	76.9	22.5	[61.4, 92.5]
HI (ohm/s2)	6.9	1.4	[5.7, 8.0]	10.1	3.3	[7.8, 12.4]
Trauma
HR (bpm)	71.6	5.4	[67.1, 76.1]	70.1	13.2	[61.0, 79.3]
RMSSD (ms)	26.7	20.8	[9.3, 44.0]	36.2	26.0	[18.2, 54.2]
SBP (mm Hg)	145.5	17.3	[131.0, 160.0]	154.2	30.7	[132.9, 175.4]
DBP (mm Hg)	80.4	16.6	[66.5, 94.2]	82.8	10.8	[75.4, 90.3]
TPR (mmHg.min/L)	5.2	1.7	[3.8, 6.6]	5.7	1.3	[4.8, 6.6]
PEP (ms)	64.6	18.1	[49.5, 79.8]	72.7	23.8	[56.3, 89.2]
HI (ohm/s2)	6.8	1.7	[5.4, 8.2]	10.7	4.5	[7.6, 13.8]
Reactivity
HR (bpm)	1.3	2.3	[−0.6, 3.3]	3.3	4.7	[0.1, 6.6]
RMSSD (ms)	−1.5	8.1	[−8.4, 5.3]	0.9	6.2	[−3.4, 5.1]
SBP (mm Hg)	6.8	6.6	[1.3, 12.3]	6.3	15.4	[−4.4, 16.9]
DBP (mm Hg)	7.4	8.9	[3.5, 14.8]	9.0	16.8	[−2.6, 20.6]
TPR (mmHg.min/L)	0.7	1.1	[3.4, 5.6]	1.0	1.9	[−0.4, 2.3]
PEP (ms)	−5.2	5.8	[−10.0, −0.3]	−4.2	2.2	[−5.7, 2.7]
HI (ohm/s2)	−0.1	0.8	[−0.7, 0.6]	0.6	1.5	[−0.5, 1.7]

A robust association was found between PTSDSS and HI for neutral (coefficient = 0.05; CI = [0.001, 0.1]; p = 0.04), trauma (coefficient = 0.07; CI = [0.01, 0.12]; p = 0.02), and reactivity conditions (coefficient = 0.02; CI = [0.001, 0.03]; p = 0.04). No association was found between the PTSDSS and PEP for neutral (coefficient = 0.03; CI = [−0.35, 0.41]; p = 0.9), trauma (coefficient = 0.0; CI = [−0.39, 0.39]; p = 1.0) and reactivity (coefficient = −0.03; CI = [−0.1, 0.06]; p = 0.5).

Analysis of the association between the with-in pair PTSD status/PTSDSS and non-ICG-based physiological metrics is presented in Appendix B. No significant association was found between the with-in pair PTSD status and HR, RMSSD, SBP, DBP, or TPR for neutral, trauma, and reactivity as depicted in Table B7. A significant positive association was found between the PTSDSS and HR (for all stages), and DBP (trauma recall, and reactivity) (Table B8). A significant, negative association between the PTSDSS and RMSSD (neutral and trauma recall) was also observed (Table B8).

On analyzing the univariate relationship between the three-level PTSD status (current PTSD, past PTSD, no-PTSD) and non-ICG-based physiological metrics for eight PTSD-discordant twin pairs, a significant, positive association was found between the three-level PTSD status and HR (in neutral), DBP (in trauma), and TPR (in trauma and reactivity) only.

4. Discussion

In this co-twin control study of Vietnam Era veterans, we found that PTSD status was associated with lower HI and PEP. These associations were generally exacerbated by the trauma recall challenge, known to elicit PTSD symptoms acutely. These associations were robust and persisted after accounting for sociodemographic and lifestyle factors, cardiovascular risk factors, and comorbid psychiatric conditions. This suggests both increased sympathetic activation and reduced cardiac contraction with trauma recall stress. Because similar pathophysiology is observed in heart failure, these findings raise the concern that PTSD stress activation may predispose to future heart failure risk and underscores the need for more related research.

The relevance of our findings is supported by prior research that emphasizes the novelty and importance of HI in heart failure. Laboratory experiments with animals have validated HI as a potential index of myocardial contractility (Peng et al., 2004) and positive inotropy of the heart muscle (Thayer et al., 2010; Peczalski 2021). HI was found to be a more reliable contractility metric than PEP (Peng et al., 2004). Low HI values have been observed in patients with clinical evidence of heart failure (Hubbard et al., 1986). A low HI was also correlated with ventricular wall tension, the maximum rate of pressure-rise in the ventricle (Thompson et al., 1981), and 2D fractional shortening (Jurn & Heethaar, 1999). HI may also be a reliable indicator of inter- and intra-subject changes in LV contractility (Van et al., 1999).

Our findings are particularly remarkable because PTSD was also associated with lower PEP. A lower PEP indicates increased sympathetic activation, and normally the latter would increase contractility by activation of the beta-1 adrenergic receptors. As such, HI is expected to increase rather than decrease in healthy individuals. Although low PEP and HI are associated with heart failure, we cannot prove causal relationships connecting PTSD, HI/PEP, and heart failure due to the lack of long-term heart failure outcomes. In most normal circumstances, PEP and HI are inversely correlated because sympathetic activity increases systolic excursion velocities. We found that they are concordantly related in PTSD, such that higher sympathetic activity leads to lower systolic velocities. Our findings of decreased HI despite reduced PEP in PTSD suggest that trauma recall may cause transient LV dysfunction through other mechanisms, which may include transient myocardial ischemia or other mechanical/electrophysiological effects. It is possible that these mechanisms also occur in heart failure. Unfortunately, we do not have simultaneous myocardial perfusion imaging to examine for ischemia as a possible mediator, but nonetheless, this is worthwhile to examine in future studies.

We also evaluated PTSD symptoms using the PTDSS, which allowed us to specifically examine whether the acute PTSD symptom burden during the time of the trauma recall challenge associated with changes in cardiovascular physiology. Although most of the associations did not reach statistical significance, the overall direction of associations was mostly similar except for HI reactivity. Overall, these findings suggest that acute PTSD symptoms, in addition to their overall PTSD status, may contribute to cardiovascular abnormalities as discussed, although the strength of association is likely relatively weak compared to PTSD status.

Mixed evidence of autonomic activation was also found in the form of a strong association of PTSD-status/PTSDSS with HR (Table B1, Table B7), RMSSD (Table B7), DBP (Table B5, Table B7), and TPR (Table B6). Similar mixed evidence of autonomic activation was observed via a strong association of the three-label PTSD status (Current-PTSD, Past PTSD, no-PTSD) with HR (in neutral), DBP (in trauma), and TPR (in trauma and reactivity).

This study is subject to several limitations. No a-priori power was considered when planning the study, as the main outcome of the parent study was cardiac imaging outcomes (Vaccarino et al., 2022). The findings of this study are subject to validation in larger prospective cohorts because of this limitation. The sample size of PTSD participants was small and might have limited the statistical power to detect small differences. We have attempted to increase the power of our analyses by also examining PTSD symptoms as a continuous exposure variable as a secondary analysis. Despite the small sample size, the co-twin design facilitates inference by reducing the impact of unmeasured differences/confounders between individuals with and without PTSD. Another limitation to our findings is that we are not able to examine PTSD with and without depression as separate entities. Nonetheless, since 10% of twins without PTSD also have depression, we were able to adjust for depression in our model and found similar results as the models without depression. We also adjusted for depression in models studying PTSD symptoms rather than the diagnosis, which gave us more statistical power in examining the effects of depression. The homogeneous sample of mostly White, middle-aged men limited the generalizability of the study to women and other racial / age groups. However, the internal validity of the study increased as a result of this homogeneity. Lastly, the SNS activity in participants was not measured using an independent metric such as microneurography-based sympathetic measurement. Nonetheless, PEP is a validated metric for cardiac specific sympathetic activity, which may differ from overall sympathetic activity (Newlin & Levenson, 1979).

5. Conclusion

In conclusion, we found robust relationships between PTSD and worsened cardiac contractility, as well as increased sympathetic activation, during trauma recall stress. This is the first investigation to examine the association of PTSD and objective measures of ICG-based contractility metrics during a trauma recall and may help understand the mechanisms through which PTSD may increase the downstream heart failure risk. More work, however, is needed to understand cardiac risk in PTSD in order to develop interventions for heart failure risk reduction.

Appendix A: Comparison between Three-Level PTSD Status

This appendix presents a comparison of characteristics between current PTSD, past PTSD, and non-PTSD participants (Table A1), and descriptive statistics (Table A2) for physiological metrics for three-level PTSD status in the neutral condition, trauma recall, and reactivity.

Table A1.

Sociodemographic, lifestyle, cardiovascular risk factors, psychiatric diagnosis, and respective medications by three-label PTSD status (current PTSD. Past PTSD and Non-PTSD).²

Factors	Current PTSD (N = 17)	Past PTSD (N = 15)	Non-PTSD (N = 105)
Sociodemographic factors
Age, years, mean (SD)	67.8 (1.6)	68.1 (2.7)	68.6 (2.6)
Non-white, %	12	0	5
Married, %	59	60	82
Years of education, mean (SD)	12.3 (1.4)	13.8 (2.1)	14.2 (2.5)
Employed full time, %	18	20	24
Service in Southeast Asia, %	65	53	33
Lifestyle factors
Cigarette Smoking
Never, %	24	7	40
Former, %	53	66	49
Current, %	24	27	11
Physical Activity (Baecke Score), mean (SD)	7.6 (1.1)	7.7 (1.4)	8.1 (1.3)
Cardiovascular risk factors and medications
BMI, mean (SD)	31.7 (5.1)	28.5 (3.9)	29.3 (4.2)
SBP, mm Hg, mean (SD)	139.7 (13.5)	139.8 (24.7)	141.1 (16.3)
DBP, mm Hg, mean (SD)	81.8 (15.3)	77.1 (9.6)	80.4 (10.8)
History of hyperlipidemia, %	65	60	67
History of diabetes, %	30	13	24
History of hypertension, %	47	53	54
Self-reported coronary heart disease (CHD), %	6	27	14
Perfusion defects >5% in size on rest or stress cardiac PET, %	42	13	18
Aspirin, %	24	27	51
Statins, %	35	47	63
Beta-blockers, %	18	27	24
ACE Inhibitors, %	6	20	27
Psychiatric diagnoses (lifetime) and medications
Major Depressive disorder, %	82	27	8
Alcohol Abuse (with or without dependence), %	47	33	21
Drug Abuse (with or without dependence), %	24	13	11
PTSD Severity score, mean (SD)	34.6 (12.0)	23.1 (9.6)	21.3 (8.1)
Antidepressants, %	65	47	6

²SD = Standard deviation; mm = millimeter; Hg = Mercury; PET = Positron Emission Tomography; ACE = Angiotensin Converting Enzyme.

Table A2:

Descriptive statistics for heart rate (HR) in beats per minute (bpm), root mean square of successive difference (RMSSD) in milli-second (ms), systolic blood pressure (SBP) in milli-meter of mercury (mm Hg), diastolic blood pressure (DBP) in mm Hg, total peripheral resistance (TPR) in mmHg.min/L, pre-ejection period (PEP) in ms, and Heather index (HI) in ohm/seconds² (ohm/s²) for three-level PTSD status (current PTSD (n =17), past PTSD (n =15), and non-PTSD participants (n =105)) in neutral, trauma, and reactivity. The HR measurements were derived from the ECG signal over 60-seconds non-overlapping sampling interval. The BP (SBP and DBP) measurements were taken twice during the neutral and trauma sessions. Then, the BP and HR measurements were averaged for each session separately. PEP and HI were also computed over 60-seconds non-overlapping windows and averaged for each session separately. The reactivity metrics (trauma minus neutral) for each participant were computed by subtracting the averaged neutral session metrics from the averaged trauma session metrics.

Metrics	Current PTSD (17)			Past PTSD (15)			No PTSD (120)
	Mean	SD	95% CI	Mean	SD	95 % CI	Mean	SD	95% CI
Neutral
HR (bpm)	72.3	11.6	[66.3, 78.3]	71.9	11.4	[65.6, 78.2]	66.5	9.4	[64.7, 68.4]
RMSSD (ms)	35.4	33.6	[19.4, 51.4]	24.7	23.4	[12.0, 37.4]	29.6	29.2	[24.0, 35.2]
SBP (mm Hg)	146.8	20.0	[136.5, 157.1]	144.9	38.0	[123.9, 165.9]	145.8	21.9	[141.6, 150.1]
DBP (mm Hg)	79.6	15.8	[71.5, 87.7]	76.7	16.2	[67.7, 85.7]	80.3	13.2	[77.8, 82.9]
TPR (mmHg.min/L)	5.1	1.6	[4.3, 5.9]	5.0	1.7	[4.0, 5.9]	5.6	1.6	[5.3, 5.9]
PEP (ms)	68.4	21.8	[57.2, 79.6]	66.9	18.7	[56.5, 77.3]	73.1	21.3	[69.0, 77.3]
HI (ohm/s2)	7.0	2.2	[5.8, 8.1]	9.7	2.3	[8.4, 11.0]	8.9	2.8	[8.3, 9.4]
Trauma
HR (bpm)	75.3	12.8	[68.7, 81.9]	73.8	12.6	[66.8, 80.8]	69.4	10.8	[67.3, 71.5]
RMSSD (ms)	34.4	29.8	[20.3, 48.6]	23.4	23.8	[10.5, 36.3]	28.2	27.2	[22.9, 33.4]
SBP (mm Hg)	151.5	16.6	[142.9, 160.1]	151.2	39.5	[129.3, 173.1]	154.2	23.3	[149.6, 158.7]
DBP (mm Hg)	86.9	17.5	[77.8, 95.9]	82.6	16.3	[73.6, 91.6]	86.6	13.3	[84.0, 89.1]
TPR (mmHg.min/L)	5.8	1.8	[4.9, 6.7]	5.5	1.8	[4.5, 6.5]	6.0	1.6	[5.7, 6.3]
PEP (ms)	60.9	18.7	[51.3, 70.6]	67.1	20.0	[56.0, 78.1]	86.6	13.3	[84.0, 89.1]
HI (ohm/s2)	7.2	2.4	[6.0, 8.4]	9.8	2.6	[8.4, 11.2]	9.2	3.1	[8.6, 9.8]
Reactivity
HR (bpm)	3.0	4.6	[0.6, 5.4]	2.0	2.8	[0.4, 3.5]	2.9	4.7	[2.0, 3.8]
RMSSD (ms)	−1.0	11.5	[−6.5, 4.5]	−1.3	4.7	[−3.9, 1.2]	−1.5	7.6	[−2.9, 4.3]
SBP (mm Hg)	4.7	13.6	[−2.3, 11.7]	6.3	7.7	[2.1, 10.6]	8.3	14.7	[5.5, 11.2]
DBP (mm Hg)	7.3	9.4	[2.4, 12.1]	5.9	12.1	[−0.8, 12.6]	6.2	8.6	[4.5, 7.9]
TPR (mmHg.min/L)	0.7	0.9	[0.2, 1.1]	0.5	1.4	[−0.2, 1.3]	0.3	0.9	[0.2, 0.5]
PEP (ms)	−7.4	8.5	[−11.8, −3.1]	0.2	7.1	[−3.7, 4.1]	−2.3	5.7	[−3.4, −1.2]
HI (ohm/s2)	0.2	0.8	[−0.2, 0.6]	0.1	0.8	[−0.3, 0.5]	0.3	0.9	[0.1, 0.5]

Appendix B: Analysis of the Association between PTSD Status and non-ICG-based Metrics.

This appendix presents the results of the analysis of the association between PTSD status /PTSDSS for PTSD as individuals as well as within-pair and non-ICG-based metrics.

Table B1:

Multivariable analysis of the relationship between PTSD status and HRV metrics (Heart rate (HR) and root mean square of successive differences (RMSSD)) for PTSD participants as individuals.

Model	Neutral		Trauma		Reactivity
	Coefficient	[95 % CI]	Coefficient	[95 % CI]	Coefficient	[95 % CI]
Outcome: HR
Model 1a	4.5	[−0.5, 9.6]	4.8	[−0.97, 10.6]	0.06	[−2.2, 2.4]
Model 2b	3.8	[−1.3, 8.9]	3.8	[−2.0, 9.6]	−0.06	[−2.4, 2.3]
Model 3c	5.3	[0.01, 10.6]	5.6	[−0.4, 11.7]	0.43	[−2.1, 2.9]
Model 4d	6.9	[0.5, 13.3]	7.3	[−0.01, 14.7]	0.64	[−2.4, 3.7]
Outcome: RMSSD
Model 1a	−0.96	[−14.6, 12.6]	1.08	[−12.4, 14.6]	0.37	[−3.8, 4.5]
Model 2b	1.02	[−12.9, 15.0]	2.1	[−11.6, 15.9]	0.82	[−3.4, 5.1]
Model 3c	8.4	[−5.9,22.7]	6.8	[−7.6, 21.3]	−2.0	[−6.4, 2.4]
Model 4d	9.9	[−7.1, 26.8]	8.1	[−9.4, 25.7]	−3.4	[−8.6, 1.9]

^aModel 1 = Unadjusted base model for PTSD status

^bModel 2 = Model 1 + sociodemographic and lifestyle factors including marital status, employment, service in Southeast Asia, current smoking status, and physical activity.

^cModel 3 = Model 2 + cardiovascular risk factors including BMI, diabetics, hypertension, history of CHD, cardiac PET-based perfusion defects, and current use of aspirin, statin, beta-blocker, and angiotensin converting enzyme.

^dModel 4 = Model 3 + psychiatric diagnosis including lifetime history of major depressive disorder, substance abuse, and current use of antidepressants.

Table B2:

Multivariable analysis of the relationship between PTSD status and BP metrics (Systolic blood pressure (SBP) and Diastolic blood pressure (DBP)) for PTSD participants as individuals.

Model	Neutral		Trauma		Reactivity
	Coefficient	[95 % CI]	Coefficient	[95 % CI]	Coefficient	[95 % CI]
Outcome: SBP
Model 1a	1.03	[−11.3, 13.4]	−2.47	[−15.2, 10.3]	−2.97	[−10.2, 4.3]
Model 2b	0.92	[−11.6, 13.4]	−0.53	[−13.5, 12.5]	−0.62	[−8.0, 6.7]
Model 3c	−4.0	[−16.8, 8.8]	−6.1	[−19.4, 7.2]	−1.6	[−9.3, 6.1]
Model 4d	−3.4	[−19.2, 12.4]	−3.7	[−20.1, 12.7]	0.6	[−8.8, 9.9]
Outcome: DBP
Model 1a	−0.22	[−7.4, 6.9]	0.6	[−6.6, 7.8]	1.09	[−3.6, 5.8]
Model 2b	0.38	[−7.0, 7.8]	1.2	[−6.2, 8.5]	1.5	[−3.4, 6.3]
Model 3c	−0.61	[−8.3, 7.1]	0.2	[−7.7, 8.1]	1.0	[−4.2, 6.1]
Model 4d	−1.2	[−10.5, 8.2]	−0.03	[−9.6, 9.6]	1.6	[−4.7, 7.9]

^aModel 1 = Unadjusted base model for PTSD status

^bModel 2 = Model 1 + sociodemographic and lifestyle factors including marital status, employment, service in Southeast Asia, current smoking status, and physical activity.

^cModel 3 = Model 2 + cardiovascular risk factors including BMI, diabetics, hypertension, history of CHD, cardiac PET-based perfusion defects, and current use of aspirin, statin, beta-blocker, and angiotensin converting enzyme.

^dModel 4 = Model 3 + psychiatric diagnosis including lifetime history of major depressive disorder, substance abuse, and current use of antidepressants.

Table B3:

Multivariable analysis of the relationship between PTSD status and total peripheral resistance (TPR) for PTSD participants as individuals. TPR was computed as ratio of mean arterial pressure (MAP) and cardiac output (CO). MAP was measured as DBP + (1/3)*(SBP− DBP), whereas CO was calculated as [(SBP −DBP)/(SBP + DBP)]*HR.

Model	Neutral		Trauma		Reactivity
	Coefficient	[95 % CI]	Coefficient	[95 % CI]	Coefficient	[95 % CI]
Outcome: TPR
Model 1a	−0.4	[−1.2, 0.47]	−0.1	[−0.94, 0.72]	0.3	[−0.16, 0.83]
Model 2b	−0.22	[−1.09, 0.65]	−0.02	[−0.87, 0.82]	0.28	[−0.24, 0.80]
Model 3c	−0.28	[−1.24, 0.68]	−0.14	[−1.1, 0.79]	0.19	[−0.36, 0.74]
Model 4d	−0.57	[−1.7, 0.57]	−0.47	[−1.6, 0.63]	0.22	[−0.46, 0.90]

^aModel 1 = Unadjusted base model for PTSD status

^bModel 2 = Model 1 + sociodemographic and lifestyle factors including marital status, employment, service in Southeast Asia, current smoking status, and physical activity.

^cModel 3 = Model 2 + cardiovascular risk factors including BMI, diabetics, hypertension, history of CHD, cardiac PET-based perfusion defects, and current use of aspirin, statin, beta-blocker, and angiotensin converting enzyme.

^dModel 4 = Model 3 + psychiatric diagnosis including lifetime history of major depressive disorder, substance abuse, and current use of antidepressants.

Table B4:

Multivariable analysis of the relationship between PTSDSS and HRV metrics (Heart rate (HR) and root mean square of successive differences (RMSSD)) for PTSD participants as individuals. PTSDSS was significantly associated with HR in the neutral and trauma stages. PTSDSS was not associated with RMSSD.

Model	Neutral		Trauma		Reactivity
	Coefficient	[95 % CI]	Coefficient	[95 % CI]	Coefficient	[95 % CI]
Outcome: HR
Model 1a	0.22	[0.04, 0.39]	0.27	[0.07, 0.46]	0.05	[−0.03, 0.13]
Model 2b	0.21	[0.04, 0.38]	0.26	[0.07, 0.45]	0.05	[−0.02, 0.13]
Model 3c	0.29	[0.11, 0.46]	0.35	[0.16, 0.55]	0.06	[−0.02, 0.15]
Model 4d	0.35	[0.16, 0.52]	0.41	[0.21, 0.61]	0.07	[−0.02, 0.16]
Outcome: RMSSD
Model 1a	−0.41	[−0.88, 0.06]	−0.36	[−0.82, 0.11]	0.02	[−0.13, 0.16]
Model 2b	−0.36	[−0.84, 0.12]	−0.32	[−0.80, 0.15]	0.03	[−0.12, 0.17]
Model 3c	−0.31	[0.82, 0.19]	−0.29	[−0.79, 0.21]	0.004	[−0.14, 0.15]
Model 4d	−0.37	[−0.90, 0.17]	−0.32	[−0.85, 0.21]	0.01	[−0.15, 0.17]

^aModel 1 = Unadjusted base model for PTSDSS

^bModel 2 = Model 1 + sociodemographic and lifestyle factors including marital status, employment, service in Southeast Asia, current smoking status, and physical activity.

^cModel 3 = Model 2 + cardiovascular risk factors including BMI, diabetics, hypertension, history of CHD, cardiac PET-based perfusion defects, and current use of aspirin, statin, beta-blocker, and angiotensin converting enzyme.

^dModel 4 = Model 3 + psychiatric diagnosis including lifetime history of major depressive disorder, substance abuse, and current use of antidepressants.

Table B5:

Multivariable analysis of the relationship between PTSDSS and BP metrics (Systolic blood pressure (SBP) and Diastolic blood pressure (DBP)) for PTSD participants as individuals. PTSDSS was significantly associated with DBP in the neutral and trauma stages. PTSDSS was not associated with SBP.

Model	Neutral		Trauma		Reactivity
	Coefficient	[95 % CI]	Coefficient	[95 % CI]	Coefficient	[95 % CI]
Outcome: SBP
Model 1a	0.14	[−0.28, 0.56]	0.01	[−0.42, 0.45]	−0.14	[−0.39, 0.10]
Model 2b	0.17	[−0.26, 0.59]	0.09	[−0.35, 0.53]	−0.08	[−0.33, 0.17]
Model 3c	−0.08	[−0.50, 0.34]	−0.21	[−0.65, 0.23]]	−0.16	[−0.41, 0.09]
Model 4d	−0.07	[−0.52, 0.37]	−0.14	[−0.61, 0.33]	−0.1	[−0.37, 0.16]
Outcome: DBP
Model 1a	0.29	[0.05, 0.52]	0.33	[0.08, 0.57]	0.03	[−0.13, 0.19]
Model 2b	0.32	[0.08, 0.57]	0.36	[0.11, 0.60]	0.04	[−0.12, 0.21]
Model 3c	0.30	[0.05, 0.55]	0.36	[0.10, 0.62]	0.05	[−0.12, 0.22]
Model 4d	0.33	[0.07, 0.59]	0.41	[0.14, 0.68]	0.08	[−0.09, 0.26]

^aModel 1 = Unadjusted base model for PTSDSS

^bModel 2 = Model 1 + sociodemographic and lifestyle factors including marital status, employment, service in Southeast Asia, current smoking status, and physical activity.

^cModel 3 = Model 2 + cardiovascular risk factors including BMI, diabetics, hypertension, history of CHD, cardiac PET-based perfusion defects, and current use of aspirin, statin, beta-blocker, and angiotensin converting enzyme.

^dModel 4 = Model 3 + psychiatric diagnosis including lifetime history of major depressive disorder, substance abuse, and current use of antidepressants.

TABLE B6.

Multivariable analysis of the relationship between PTSDSS and total peripheral resistance (TPR) for PTSD participants as individuals

Model	Neutral		Trauma		Reactivity
	Coefficient	95% CI	Coefficient	95% CI	Coefficient	95% CI
Outcome: TPR
Model 1 a	0.02	[−0.01, 0.04]	0.03	[−0.001, 0.06]	0.01	[−0.01, 0.03]
Model 2 b	0.02	[−0.01, 0.05]	0.03	[0.0004, 0.06]	0.01	[−0.01, 0.03]
Model 3 c	−0.02	[−0.01, 0.05]	0.03	[0.0008, 0.06]	0.01	[−0.004, 0.03]
Model 4 d	−0.01	[−0.02, 0.05]	0.03	[−0.005, 0.06]	0.02	[−0.002, 0.04]

Note: TPR was computed as the ratio of mean arterial pressure (MAP) and cardiac output (CO). MAP was measured as DBP + (1/3)*(SBP − DBP). CO was measured as [(SBP − DBP)/(SBP + DBP)]*HR. PTSDSS was significantly associated with TPR in trauma for Model 2 and Model 3 only.

^aModel 1 = Unadjusted base model for PTSDSS.

^bModel 2 = Model 1 + sociodemographic and lifestyle factors including marital status, employment, service in Southeast Asia, current smoking status, and physical activity.

^cModel 3 = Model 2 + cardiovascular risk factors including BMI, diabetics, hypertension, history of CHD, cardiac PET-based perfusion defects, and current use of aspirin, statin, beta-blocker, and angiotensin converting enzyme.

^dModel 4 = Model 3 + psychiatric diagnosis including lifetime history of major depressive disorder, substance abuse, and current use of antidepressants.

Table B7:

Univariate analysis of relationship between PTSD status and HR, RMSSD, SBP, DBP, and TPR for PTSD-discordant pairs, comparing in neutral, trauma, and reactivity. No significant association was found between PTSD status and any of these physiological metrics for any stage.

Outcome	Neutral		Trauma		Reactivity
	Coefficient	[95 % CI]	Coefficient	[95 % CI]	Coefficient	[95 % CI]
HR	3.5	[−1.5, 8.4]	1.5	[−4.8, 7.8]	−1.9	[−4.4, 0.38]
RMSSD	−7.1	[−17.5, 3.3]	−9.5	[−20.8, 1.8]	−2.4	[−6.1, 1.3]
SBP	−9.3	[−23.9, 5.4]	−8.7	[−22.5, 5.2]	0.6	[−6.9, 8.0]
DBP	−0.81	[−8.5, 6.9]	−2.4	[−9.5, 4.6]	−1.6	[−7.2, 3.9]
TPR	−0.22	[−1.1, 0.69]	−0.46	[−1.3, 0.39]	−0.24	[−0.9, 0.4]

Table B8:

Univariate analysis of the relationship between PTSDSS and non-ICG based physiologica metrics (HR, RMSSD, SBP, DBP, and TPR) for PTSD-discordant pairs, comparing in neutral, trauma, and reactivity. A significant association was found between PTSDSS and HR (for all stages), RMSSD (neutral and trauma recall), and DBP (trauma recall, and reactivity).

Outcome	Neutral		Trauma		Reactivity
	Coefficient	[95 % CI]	Coefficient	[95 % CI]	Coefficient	[95 % CI]
HR	0.2	[0.03, 0.38]	0.31	[0.09, 0.5]	0.1	[0.02, 0.19]
RMSSD	−0.5	[−0.86, −0.11]	−0.47	[−0.88, −0.06]	0.01	[−0.13, 0.15]
SBP	0.28	[−0.26, 0.81]	0.34	[−0.16, 0.85]	0.07	[−0.2, 0.34]
DBP	0.19	[−0.08, 0.47]	0.42	[0.18, 0.67]	0.22	[0.02, 0.43]
TPR	−0.02	[−0.05, 0.01]	0.0002	[−0.03, 0.03]	0.02	[−0.004, 0.04]