Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Jan 1.
Published in final edited form as: Biol Psychol. 2022 Dec 6;176:108472. doi: 10.1016/j.biopsycho.2022.108472

Central and Peripheral Nervous System Responses to Chronic and Paced Hyperventilation in Anxious and Healthy Subjects

David F Tolin a,b, Emily M O’Bryan a,c, Carolyn D Davies a,d, Gretchen J Diefenbach a,b, Jason Johannesen b
PMCID: PMC9839632  NIHMSID: NIHMS1856230  PMID: 36481266

Abstract

The aim of the present study was to examine self-report, peripheral nervous system, and central nervous system correlates of naturally-occurring, chronic hyperventilation (HV, assessed by hypocapnia or low resting state low end-tidal CO2), and to examine the additional effect of acute, experimentally-induced HV in anxious and healthy participants. By identifying the biomarkers of anxiety-related chronic HV and examining responses to acute HV, we hope to identify meaningful, mechanistic targets for further treatment development. Seventy anxious patients and 34 healthy control participants completed electroencephalogram (EEG) and peripheral nervous system recording at baseline and following a paced breathing task. Diagnosis x baseline hypnocapnia group analyses indicated that anxious/hypocapnic patients exhibited greater nonspecific skin conductance response amplitude than did anxious/normocapnic patients, and the anxious group reported greater HV-related symptoms and anxiety sensitivity than did the control group. However, no EEG abnormalities were noted as a function of anxiety group or baseline hypocapnia status. Following paced HV, anxious patients (but not controls) exhibited an increase in left-frontal alpha 1 power. Hypocapnic, but not normocapnic, participants exhibited an increase in skin conductance levels. Anxious patients reported an increase in negative cognitive appraisals of HV symptoms, and anxious/hypocapnic participants reported an increase in affective responses to HV. Thus, chronic HV is associated with greater arousal, and increased self-reported and physiological sensitivity to paced HV. Patients who chronically hyperventilate appear to be more sensitive to respiratory distress, responding with higher levels of anxiety and poorer tolerance of the physiological sensations accompanying acute HV.

Keywords: EEG, psychophysiology, anxiety

The peripheral nervous system (PNS) biomarker of hyperventilation (HV) is low arterial CO2 (hypocapnia), commonly measured via end-tidal CO2 (ETCO2). HV rapidly decreases arterial partial pressure of CO2 (PaCO2). As PaCO2 levels change, the body seeks homeostasis through compensatory mechanisms. HV-related hypocapnia initiates a cascade of physical symptoms such as lightheadedness, increased heart rate, and muscle weakness, resulting from vasoconstriction and impaired cerebral blood flow, and contractility changes in smooth muscle (Gilbert, 1999). These physiological states of HV are often associated with feelings of anxiety, and in some cases may prompt panic attacks resulting from a fear of fear cycle (anxiety further escalates the physical symptoms thus further escalating anxiety). The fear-of-fear cycle may be particularly pronounced in individuals with high anxiety sensitivity, or a tendency to misinterpret symptoms, such as those associated with HV, as dangerous. Acute HV occurs abruptly when respiration is suddenly and distinctly faster and deeper from one’s usual breathing pattern. Chronic HV may also occur when an individual’s usual breathing pattern routinely leads to the expulsion of excessive levels of CO2 producing a chronic physiological state of hypocapnia. Unlike acute HV, which causes abrupt changes in physiology, states of chronic hypocapnia are associated with subtle symptoms, which may not even be perceived by the individual.

Given the association with panic attacks, HV has been studied extensively in patients with panic disorder (Klein, 1993; Ley, 1985). Respiratory characteristics of patients with panic disorder include increased frequency and depth of sigh breaths (Wilhelm et al., 2001), and increased respiratory variability preceding panic attacks (Meuret et al., 2011; Rosenfield et al., 2010). Panic patients exhibit significantly lower ETCO2 values (indicative of chronic HV) than do healthy control subjects (Hegel & Ferguson, 1997; Munjack et al., 1993; Rapee, 1986). Increasingly, however, research shows that chronic HV, and the resulting physiological state of hypocapnia, is not limited to panic disorder, and may cut across anxiety disorder diagnoses: Multiple studies have found equivalent rates of chronic hypocapnia in patients with panic vs. other anxiety disorders (Davies & Craske, 2014; Tolin, Billingsley, et al., 2017; Tolin, McGrath, et al., 2017; van den Hout et al., 1992).

For patients with chronic HV, there is good reason to suspect that the resulting hypocapnia serves to maintain the illness and inhibit treatment response. The presence of hypocapnia and its attendant unpleasant physiological sensations may increase anxiety sensitivity (e.g., catastrophic misappraisals of interoceptive stimuli) (Ley, 1985) and stimulate further HV (Dempsey et al., 1975; Smoller et al., 1996), thus creating a self-perpetuating cycle. Cognitive-behavioral therapy is an empirically-supported treatment for anxiety disorders (e.g., Hofmann & Smits, 2008). However, anxious patients with chronic hypocapnia show a demonstrably poorer response to cognitive-behavioral therapy (CBT). In one study of CBT for a range of anxious patients, lower baseline ETCO2 significantly predicted poorer outcomes on a standardized measure of anxiety symptoms (Davies & Craske, 2014). Another study found a significant relationship between baseline ETCO2 and CBT completion in a naturalistic treatment setting, with lower baseline ETCO2 associated with an increase in the odds of premature dropout (Tolin, Billingsley, et al., 2017). Indeed, chronically hypocapnic patients were more likely to drop out than they were to complete treatment, with a 70% greater likelihood of dropout compared to normocapnic patients. Thus, chronic hypocapnia (indicative of chronic HV) is a negative predictor of outcomes for otherwise effective treatments, making this a highly promising candidate target for improving treatment efficacy. Understanding the physiological and behavioral correlates of hypocapnia associated with both chronic and acute HV in patients with and without anxiety disorders is an essential step toward this goal.

As noted above, hypocapnia is associated with well-established PNS responses; however, less is known about PNS responses within the context of anxious arousal. In addition, less is known about the central nervous system (CNS) correlates of hypocapnia. Most of the current research on HV has been conducted with healthy control subjects directed to engage in acute (paced) HV. In a study designed to demonstrate a model of reversible HV-induced ischemic hypoxia, acute HV in healthy young subjects produced electroencephalograph (EEG) power to shift from higher (alpha, beta) to lower (delta, theta) frequency bands, resembling EEG changes observed in hypobaric hypoxia (Kraaier et al., 1988). Hypobaric hypoxia is associated with disruption of CNS functions, including sleep disturbance, oxidative stress, alterations of acetylcholine neurotransmitter, hippocampal dendrite atrophy, and cognitive impairment in domains of learning and memory (Muthuraju & Pati, 2014). Kraaier and colleagues (1988) further reported a decrease in blood flow velocity of 40% relative to baseline following HV, and approximate effects on cerebral blood flow to values associated with ischemic penumbra. While these effects are considered harmless when obtained under conditions of brief HV, chronic hyperventilation associated with anxiety disorder could be hypothesized to produce more pervasive effects on CNS function. Indeed, in a study comparing chronically hypocapnic and normocapnic subjects on EEG, a parameter reflecting the ratio of low-frequency (Delta + Theta) to high-frequency (Alpha + Beta) EEG power showed a 90% elevation in low-frequency activity associated with hypocapnia (Matteo et al., 1992), though HV was not manipulated in that study. It is not yet known whether these same effects occur in anxiety-related chronic HV. ETCO2 is lower following paced HV (e.g., 20–22 mmHg) (Schubert & Drummond, 1986) than is that of hypocapnic patients in clinical samples (Tolin, Billingsley, et al., 2017; Tolin, McGrath, et al., 2017); however, chronic hypocapnia at levels observed in anxiety patients could nevertheless have deleterious effects on brain function in the long term. Therefore, there is a critical gap in current knowledge: with chronic HV common in anxiety disorders, and the CNS effects of chronic HV in anxiety patients unknown, we may be overlooking an important mechanism influencing not only treatment engagement but also residual effects on brain function following successful treatment.

Some CNS correlates of anxiety symptoms have also been established based on EEG: Panic patients exhibit diminished alpha-1 power in frontal, parietal, and occipital regions (Wise et al., 2011). As alpha is inversely related to brain activity, this decrease in alpha is interpreted to reflect increased neural activation. Although not directly investigated in anxiety disorder samples, lower baseline alpha power is associated with less efficient information processing, including a diminished capacity for learning and memory (Klimesch, 1999). Both community samples with anxious arousal (Mathersul et al., 2008) and panic patients (Wise et al., 2011) exhibit frontal alpha asymmetry characterized by elevations in left relative to right-side spectral power. Again, given the inverse relationship between alpha and neural activation, left lateralization reflects more activity in the right hemisphere compared to the left. Left frontal EEG alpha lateralization, described in over 40 studies (Nusslock et al., 2015), has been interpreted as a neurophysiological marker of reduced approach-related motivation or increased withdrawal motivation, and has been shown to correlate with higher self-reported behavioral activation sensitivity (Coan & Allen, 2003). The prominence of neurophysiological abnormalities across this literature links CNS activity (specifically lower alpha-1 power and left frontal alpha asymmetry) to the genesis of anxious arousal; however, what remains less clear is the extent to which these features represent trait- (i.e., neural compromise) or state-related (i.e., acute arousal) aspects of anxiety. Although acute (paced) HV has been linked to acute changes in EEG function in non-anxious individuals, the chronic HV and subsequent hypocapnia observed in anxious patients may lead to chronic CNS effects that are not clear at this time.

To examine trait and state aspects of HV on subjective, PNS, and CNS indices of anxiety, we sampled a group of anxious patients with and without baseline hypocapnia, as well as a group of healthy control participants, also divided in terms of baseline hypocapnia. To examine these relationships at the trait level, we focused on chronic (baseline) hypocapnia, anxiety disorder diagnoses, and their relationships with self-report measures of HV symptoms and anxiety sensitivity, PNS indices of anxious arousal, and CNS (EEG) indices that have previously been associated with anxiety. To examine state-level differences, we used paced (acute) HV in which participants hyperventilated and we examined self-report measures of cognitive and affective reactions to HV symptoms, PNS indices of arousal, and CNS markers of anxiety, as well as by PNS measures of anxious arousal including higher skin conductance level (SCL), a higher number of nonspecific skin conductance responses (NSSCRs), and greater mean NSCCR amplitude.

In an attempt to better characterize group differences between anxious and non-anxious individuals as a function of baseline (i.e., trait) hypocapnia, the first aim of the present study was to determine the relationship between chronic (baseline) hypocapnia and PNS, CNS, and self-report indices of HV and anxious responding in adults with anxiety disorders. We predicted that chronically hypocapnic patients, compared to normocapnic patients and to hypocapnic and normocapnic healthy controls, would differ across several levels of analysis, including self-report, PNS, and CNS indices (see Data Analytic Strategy below). The second aim was to examine causal links between paced HV and self-report, PNS, and CNS indices of HV and anxious responding in adults with and without anxiety disorders. We predicted that within subjects, and particularly pronounced for anxious and chronically hypocapnic participants, a shift from normal breathing to paced HV would result in an enhancement of the effects predicted above.

Method

Participants

We recruited 72 adult (age 18+) patients meeting criteria for a primary diagnosis of a DSM-5 anxiety disorder of at least moderate severity. Inclusion/exclusion criteria were selected in order to obtain a clinically representative sample that could provide informed consent. Accordingly, comorbid depressive disorder and psychiatric medications were allowed. Patients with a lifetime psychotic or bipolar disorder, substance use disorder in the past 6 months, or intellectual dysfunction (as assessed via clinical records and interview) were excluded, as well as those with serious suicidal risk. Patients who were pregnant or who reported uncontrolled respiratory illness (e.g., COPD, asthma), a history of stroke, heart attack, or cardiac surgery, or history of anoxic or traumatic brain injury with loss of consciousness > 5 minutes were excluded due to potential confounds and safety risks. We also recruited 34 healthy control participants, matched to the patient sample for age and gender. Healthy controls were not permitted to have any current or historical psychiatric diagnoses. The original intent was to recruit 80 anxious patients and 40 HCs, which would be adequate to detect a clinically meaningful effect (d = 0.60, just above the traditional threshold for a medium effect); however, enrollment was interrupted by the COVID pandemic. Participants were categorized by anxiety disorder and hypocapnic status, with hypocapnia defined as ETCO2 < 35 mmHg (Sharma & Hashmi, 2021), resulting in four groups: Anxious/hypocapnic (n = 20), anxious/normocapnic (n = 50), healthy/hypocapnic (n = 10), or healthy/normocapnic (n = 24). However, analyzed sample sizes were smaller for some variables due to missing data or values failing to meet quality assurance. Participant flow summarizing analyzed sample sizes for each variable is depicted in Figure 1. Two participants were excluded from all analyses due to missing baseline hypocapnic status as a result of equipment issues. Four participants were excluded from all EEG analyses due to concerns about data quality (total n excluded = 6). One participant was excluded from all SCL analyses (total n excluded = 3) and 14 participants were excluded from NSSCR analyses at period 1 (total n excluded = 16) and 20 at period 2 (total n excluded = 22) due to no detected skin conductance responses. The same participants were excluded from NSCCR amplitude analyses; however, one additional participant was excluded at resting period 1 (total n excluded = 17) and period 2 (total n excluded = 23) due to missing a value needed to compute amplitude.

Figure 1.

Figure 1.

Open in a new tab

Flow diagram depicting participant inclusion and exclusion across analyses. Anx = anxious patients. HC = healthy controls. Hypo= hypocapnic. Normo= normocapnic.

Measures

Diagnostic status was assessed using the Diagnostic Interview for Anxiety, Mood, and Obsessive-Compulsive and Related Disorders (DIAMOND) (Tolin et al., 2018), a validated diagnostic interview for DSM-5 disorders, which includes an assessment of suicidality and shows good to excellent test-retest and inter-rater reliability. Global illness severity was estimated using the Clinician’s Global Impression Scale-Severity (CGI-S) (Guy, 1976); a rating of 4 (“moderate”) or greater was required for the patient sample.

Self-reported HV symptoms were assessed with the Nijmegen Questionnaire (NQ) (van Dixhoorn & Duivenvoorden, 1985), a well-validated measure of 16 HV-related symptoms, including chest pain, blurred vision, dizzy spells, and paresthesias (van Dixhoorn & Folgering, 2015). Items are scored from 0 (not at all) to 4 (very often), for a score range from 0–64. Internal consistency (α) in the present sample was .91.

Anxiety sensitivity was measured using the Anxiety Sensitivity Index-3 (ASI-3) (Taylor et al., 2007), a self-report measure that assesses physical, cognitive, and social concerns related to anxious sensations and correlates significantly with anxious response to biological challenge (Carter et al., 2009; Carter et al., 2001). Internal consistency (α) in the present sample was .93. To measure acute reactions following the paced HV condition, we administered the brief version of the Hyperventilation Questionnaire (HVQ-B) (Sabourin et al., 2013), a self-report measure of cognitive (e.g., “feeling trapped or helpless”), and affective (e.g., “fear”), and somatic (e.g., “dizziness”) responses to arousal induction exercises. Of particular interest here are the cognitive and affective subscales, which show excellent internal consistency and construct validity (Sabourin et al., 2013). Internal consistency (α) in the present sample was .53 for the cognitive subscale, and .83 for the affective subscale.

Apparatus

Hypocapnia was assessed via ETCO2 with the BIOPAC MP150 system, a modular data acquisition and analysis system that is widely used in psychophysiological research. ETCO2 is collected via nasal cannula and is measured in real time using the AcqKnowledge software rate calculation function, which measures the peak of the CO2 signal to provide a breath-by-breath measurement of percent ETCO2 which can be statistically transformed to mmHg.

EEG was recorded using the MP150 bio-amplifier coupled with BioNomadix wireless receiver modules. Six passive Ag/AgCl sintered disk electrodes were affixed to neoprene caps (EASYCAP GmbH) at sites two frontal (F3, F4), two midline (FZ, PZ), and two occipital (O1, O2) scalp sites according to the 10–20 system. Additional electrodes were placed bilaterally at mastoids for reference, and the outer canthi of both eyes (horizontal electrooculogram; HEOG), and above and below the right orbit (vertical electrooculogram; VEOG) for measurement of ocular artifact. Continuous EEG was monitored online in AcqKnowledge software and acquired at a 512 Hz sampling rate. Experiment timing and stimulus delivery were administered by Neurobehavioral Systems (NBS) Presentation software, with behavioral responses captured using labeled buttons on a computer keyboard. EEG signal processing was conducted in Brain Vision Analyzer v2.1 software.To measure the extent to which chronic and acute HV are associated with tonic or phasic changes in sympathetic nervous system activity, we measured electrodermal activity (EDA). EDA is highly sensitive to anxious arousal, and is considered the best direct measure of sympathetic activity. Skin conductance level (SCL), long considered the “gold standard” of resting-state anxious arousal (Dawson et al., 2007; Lykken & Venables, 1971), was recorded with a constant voltage of 0.5 V across two electrodes placed on the palmar surface of the middle phalanges of the first and second fingers of the nondominant hand. Ag/AgCl electrodes with a circular contact area of 6 mm were filled with an isotonic gel. We extracted mean SCL, the number of NCCSRs per minute, and the mean NSSCR amplitude. NSSSR frequency (or the number of SCRs in the absence of an identifiable eliciting stimulus) and amplitude (conductance at the peak relative to the conductance at the onset) represent another EDA measure of sympathetic activation; jointly, SCL and NSSCRs appear to reflect experienced emotion (Kreibig et al., 2007) and are associated with self-report of emotional arousal (e.g., Sato et al., 2020). NSSCRs, unlike SCL, may reflect more transient emotional experiences, as negative emotion ratings appear to precede NSSCRs (Gonzalez-Bono et al., 2002; Nikula, 1991).

Procedure

All procedures were approved by the hospital’s IRB, and participants provided informed consent before any study procedures took place. Anxious patients were recruited from the existing flow in a specialty outpatient clinic. Healthy control participants were primarily recruited from a research registry, with some recruited via newspaper and online advertising. Participants were reimbursed $50 for their time.

After consenting, participants were affixed with physiological recording apparatus as described above, and completed self-report measures (NQ, ASI-3). The physiological recording session consisted of collecting EEG, EDA, and ETCO2 during two separate periods. In period 1 baseline/resting state physiological measures were recorded under conditions of normal breathing. In Period 2, physiological recordings were collected after participants engaged an experimental paced HV task.

Period 1 (Normal Breathing).

During Period 1, participants were not instructed to manipulate their breathing in any way. To obtain resting state physiological assessments, participants sat quietly while focusing their eyes on a cross hair centered on the computer monitor. Participants were instructed to alternate twice between resting with eyes open and resting with eyes closed for one minute each for a total of 4 minutes. When resting with eyes open participants were cued with a message on the computer screen to “close eyes.” When resting with eyes closed participants were cued to open their eyes with an auditory tone.

Period 2 (Paced HV Task).

Period 2 (paced HV) began with a 60-second paced HV trial in which breathing rate was timed to an auditory tone at a rate of 30 breaths per minute. Hypocapnia was verified by ETCO2 and the HV session was extended as needed to obtain ETCO2 < 35 mmHg. A 4-minute resting period was completed immediately following the paced HV in order to obtain post-HV EEG and EDA recordings, and the HVQ-B was administered.

EEG Signal Processing

The resting EEG data record was filtered (Butterworth zero phase, 0.1–100 Hz, with a notch filter at 60Hz) and re-referenced to the average of 6 scalp EEG channels. Periods with low, invariant, activity (< 10uV over 1000ms interval), large voltage shifts (150 uV/ms and minimal/maximal amplitude + 75uV), and discontinuities in signal were marked (+ 300ms) according to parameters set in an automated data inspector routine. EEG was segmented first by condition (REO, REC) and then into equally-sized segments of approximately 2s each. Segments containing intervals marked for contamination were removed from further processing. Remaining segments were corrected for ocular artifact using an Infomax Restricted Independent Component Analysis (ICA). This procedure was selected after comparison with a regression-based blink correction, which poorly characterize the spatial distribution of blink artifact due, presumably, to limited spatial coverage of the EEG montage. Spectral power (μV2) was extracted by Fast Fourier Transform using a Hanning window (10% length, resolution 0.49 Hz) in the following frequency bands. Spectral power was then extracted for analysis in the following bands: theta (4–7.5 Hz), alpha-1 (8–11 Hz), alpha-2 (11–13 Hz), and beta (14.5–30 Hz). Following the approach from previous HV research (Matteo et al., 1992), a derived score found most sensitive to the effects of hypocapnia was computed based on relative power in the delta and theta bands divided by alpha and beta bands [(D+T)/(A+B)]. We also focused on the alpha-1 (8–11 Hz) band; decrease in alpha-1 power represents a state of heightened alertness (Klimesch, 1999) and has been demonstrated in patients with panic disorder (Wise et al., 2011). Frontal alpha power was evaluated as the average of alpha 1 power across electrodes F3 and F4. Frontal alpha asymmetry was computed as right (F4) alpha 1 power – left (F3) alpha 1 power. Accordingly, negative values are interpreted to suggest the presence of anxious apprehension, whereas positive values are interpreted to indicate anxious arousal.

Data Analytic Strategy

The between-subjects dependent variables studied for period 1 (normal breathing) were:

  • Self-report indices: NQ and ASI-3. We predicted that anxious/hypocapnic participants would report more HV symptoms and greater anxiety sensitivity on these measures than would the other three groups.

  • PNS indices: SCL and NSSCR frequency and amplitude. We predicted that anxious/hypocapnic participants would exhibit greater physiological arousal on these measures than would the other three groups.

  • CNS indices: The EEG parameter [(D+T)/(A+B)], resting-state frontal alpha band power, and hemispheric lateralization of frontal resting-state alpha band power. We predicted that anxious/hypocapnic participants would show more lower-frequency signal, lower alpha, and greater left alpha lateralization than would the other three groups.

For the period 1 (normal breathing) analyses, we evaluated main effects of anxiety (anxious vs. healthy) and baseline hypocapnia (hypocapnic vs. normocapnic) and their interaction across endpoints using between-groups GLM followed by oneway ANOVAs and pairwise t-tests if the interaction or main effects were significant. A statistically significant Anxiety X Hypocapnic interaction, with the anxiety/hypocapnic group differing significantly from the other three groups in post-hoc analyses, would support our hypotheses.

The within- and between-subjects dependent variables studied across period 1 (normal breathing) and period 2 (paced HV) were:

  • Self-report index: HVQ-B. We predicted a 3-way interaction in which a shift from normal breathing to paced HV would result in greater cognitive and affective reactions to HV symptoms, and that this finding would be particularly pronounced for anxious/hypocapnic participants.

  • PNS indices: SCL and NSSCR frequency and amplitude. We predicted a 3-way interaction in which a shift from normal breathing to paced HV would result in greater physiological arousal, and that this finding would be particularly pronounced for anxious/hypocapnic participants.

  • CNS indices: [(D+T)/(A+B)], resting-state frontal alpha band power, and hemispheric lateralization of frontal resting-state alpha band power. We predicted a 3-way interaction in which a shift from normal breathing to paced HV would result in lower-frequency signal, lower alpha, and greater left alpha lateralization, and that this finding would be particularly pronounced for anxious/hypocapnic participants.

For the period 1 (normal breathing) to period 2 (paced breathing) analyses, main effects of anxiety (anxious/healthy), baseline hypocapnia (hypocapnic/normocapnic) and study period (normal breathing vs. paced HV) and their interactions were evaluated using mixed-factor GLM, with period as the repeated measure. Significant omnibus tests were further examined using repeated-measures ANOVAs and t-tests. A main effect of study period would support an effect of the HV induction. A significant 3-way interaction (anxiety X hypnocapnic X study period), and post-hoc test demonstrating group differences, would support the hypothesis that the effect of the HV induction was strongest in the anxiety/hypocapnic group.

Results

Sample description

Demographic information for the four groups can be seen in Table 1. As expected, ETCO2 values differed significantly by hypocapnic status (Tukey HSD follow-up tests, p < .05). The most common diagnosis was generalized anxiety disorder and hypocapnic and normocapnic participants did not differ in terms of the frequency of any diagnosis.

Table 1.

Sample description

Healthy normocapnic (n = 24) Healthy hypocapnic (n = 10) Anxious normocapnic (n = 50) Anxious hypocapnic (n = 20) F/χ2
Age [M (SD)] 36.13 (16.15) 41.50 (18.49) 36.74 (13.89) 31.80 (12.40) 1.06
Hispanic [N (%)] 8 (33.3%) 0 (0.0%) 1 (2.0%) 3 (15.0%) 17.16**
Nonwhite [N (%)] 7 (29.2%) 2 (20.0%) 6 (12.0%) 1 (5.0%) 5.76
Panic disorder [N (%)] -- -- 11 (22.0%) 6 (30.0%) 0.48
Generalized anxiety disorder [N (%)] -- -- 30 (60.0%) 8 (40.0%) 2.30
Social anxiety disorder [N (%)] -- -- 24 (48.0%) 7 (35.0%) 0.98
Agoraphobia [N (%)] -- -- 5 (10.0%) 3 (15.0%) 0.35
Specific phobia [N (%)] -- -- 6 (12.0%) 3 (15.0%) 0.11
Separation anxiety disorder [N (%)] -- -- 1 (2.0%) 0 (0.0%) 0.41
Other specified anxiety disorder [N (%)] -- -- 4 (8.0%) 4 (20.0%) 2.03
Persistent depressive disorder [N (%)] -- -- 11 (22.0%) 5 (25.0%) 0.07
Major depressive disorder [N (%)] -- -- 17 (34.0%) 8 (40.0%) 0.22
ETCO2 [M (SD)] 39.67 (3.06) 32.43 (2.07) 39.30 (2.93) 32.40 (1.88) 47.37**
SSRI/SNRI [N (%)] -- -- 19 (59.4%) 13 (40.6%) 4.54
Open in a new tab
**

p < .01

Forty-seven percent of the anxious group was taking an SSRI/SNRI medication at the time of the study; hypocapnic and normocapnic participants did not differ significantly in terms of SSRI/SNRI use although there was a nonsignificant trend (p = 0.056) for greater SSRI/SNRI use among normocapnic participants. We examined whether outcome variables differed by SSRI/SNRI use. None of the self-report measures differed as a function of SSRI/SNRI use (p’s = .109 to .815). SCL, NSSCR frequency, and NSSCR amplitude did not differ according to SSRI/SNRI use (ps = .456, .106, and .188, respectively). For clarity, we re-ran all analyses using SSRI/SNRI use (dummy-coded as 0 or 1) as a covariate.

The EEG parameter [(D+T)/(A+B)] was significantly higher (indicating relatively greater low-frequency signal) for anxious patients taking SSRI/SNRIs than for those not taking such medications (t = 2.38, p = .02). Alpha power and alpha lateralization did not differ according to SSRI/SNRI use (ps = .098 and .359, respectively).

Period 1 (normal breathing)

Self-Report Measures.

Interaction results are shown in Table 2. On the NQ, although the anxiety x hypocapnia interaction was not significant, there was a significant main effect of anxiety, F(1, 100) = 93.55, p < .001, η2p = .483, such that mean NQ scores were higher in the anxious group relative to the healthy group, demonstrating a higher level of perceived HV symptoms among anxious participants. There was no main effect of hypocapnia (p > .05). Results did not change substantively when controlling for SSRI/SNRI use.

Table 2.

Interaction effects under normal breathing conditions.

Dependent variable Healthy normocapnic (n = 24) Healthy hypocapnic (n = 10) Anxious normocapnic (n = 50) Anxious hypocapnic (n = 20) Details Interaction F p η2p
Self-report measure of HV syndrome 4.167 (5.289) 4.000 (3.944) 20.760 (7.912) 20.050 (8.976) NQ 0.03 0.872 0.000
Anxiety sensitivity 5.875 (4.990) 3.500 (4.429) 30.440 (12.453) 25.150 (14.518) ASI-3 0.32 0.570 0.003
Mean skin conductance level (SCL) 4.244 (2.086) 4.765 (3.028) 3.954 (2.679) 3.911 (2.132) 0.24 0.623 0.002
Nonspecific skin conductance responses (NSSCRs) 6.364 (4.424) 7.600 (7.121) 8.000 (7.173) 6.765 (5.286) 0.68 0.411 0.008
Mean NSCCR amplitude 0.234 (0.158) 0.173 (0.120) 0.161 (0.147) 0.288 (0.263) 5.10 0.027 0.057
Lower- to higher-frequency EEG signal [(D+T)/(A+B)] 0.324 (0.241) 0.422 (0.147) 0.335 (0.251) 0.305 (0.106) Fz REO 1.59 0.211 0.016
Resting-state frontal alpha band power −1.894 (0.887) −1.930 (0.709) −1.950 (1.037) −2.013 (0.791) Mean of F3 and F4 REO 0.01 0.943 0.000
Left hemispheric lateralization of resting-state alpha band power (alpha 1) 0.048 (0.288) −0.101 (0.484) 0.081 (0.536) 0.142 (0.510) F4-F3 REO 0.89 0.349 0.009
Open in a new tab

Similarly, on the ASI-3, although the anxiety x hypocapnia interaction was nonsignificant, there was a significant main effect of anxiety, F(1, 100) = 81.69, p < .001, η2p = .450, such that mean ASI-3 scores were higher (demonstrating greater anxiety sensitivity) in the anxious group relative to the healthy group. There was no main effect of hypocapnia (p > .05). Results did not change substantively when controlling for SSRI/SNRI use.

Thus, anxious participants (compared to healthy controls) report higher levels of anxiety sensitivity and HV symptoms than do healthy control participants; however, this difference is unaffected by baseline hypocapnia.

PNS Psychophysiological Measures.

There was a statistically significant anxiety x hypocapnia interaction for NSSCR amplitude; examination of simple main effects revealed a significant difference in mean amplitude between anxious/hypocapnic and anxious/normocapnic participants, F(1, 85) = 6.31, p = .014, η2p = .069, such that amplitude was higher in anxious/hypocapnic participants than in anxious normocapnic participants. When controlling for SSRI/SNRI use, this interaction was no longer significant, F(1, 79) = 2.31, p = .132, η2p = .028. No additional simple main effects were found. There were no significant anxiety x hypocapnia interactions, or main effects of anxiety or hypocapnia, for mean SCL (hypothesis 4) or number of NSSCRs (hypothesis 5; all ps > .05).

Thus, NSSCR amplitude (a measure of emotional arousal) is greater among anxious patients who evidence chronic HV (e.g., are hypocapnic under conditions of normal breathing at baseline), although at present this finding is confounded with SSRI/SNRI use. We did not find similar results for other EDA measures of arousal.

EEG Measures.

There were no significant anxiety x hypocapnia interactions, or main effects of anxiety or hypocapnia, for the parameter [(D+T)/(A+B)], resting-state frontal alpha band power, or lateralization of resting-state alpha band power at alpha 1 (all ps > .05). When we included SSRI/SNRI use as a covariate in the [(D+T)/(A+B)] analysis, we similarly obtained no significant interaction effect (p = .098). There was a significant main effect of group (F = 4.21, p = .043), such that anxious participants (regardless of hypocapnic status) had lower values (indicating relatively higher EEG power in high-frequency bands) than did healthy participants.

Thus, under normal breathing conditions, anxious participants (compared to healthy control participants) may exhibit relatively higher EEG power in high-frequency bands when controlling for medication use. This effect appears unrelated to baseline hypocapnia.

Period 1 (normal breathing) to period 2 (paced HV task)

The next set of analyses examined shifts from normal breathing (period 1) to acute HV using a paced breathing task (period 2). We therefore are interested in interaction effects and main effects concerning period (see Table 3).

Table 3.

Interaction effects testing change in study outcomes from normal breathing to immediate post-hyperventilation breathing by anxiety and hypocapnia.

Dependent variable Healthy normocapnic (n = 24) Period 2 Healthy hypocapnic (n = 10) Period 2 Anxious normocapnic (n = 50) Period 2 Anxious hypocapnic (n = 20) Period 2 Details Interaction F p η2p
Cognitive responses to HV 0.083 (0.282) 0.100 (0.316) 1.500 (2.476) 2.950 (3.052) HVQ-B cognitive 2.72 0.102 0.026
Affective responses to HV 0.375 (0.770) 0.500 (0.972) 4.640 (3.421) 6.650 (4.998) HVQ-B affective 4.39 0.039 0.042
Mean SCL 4.247 (1.968) 5.045 (3.315) 3.876 (2.546) 4.329 (2.212) 0.393 0.548 0.004
Number of NSSCRs 4.714 (4.280) 4.667 (5.477) 7.108 (6.707) 5.118 (3.371) 0.471 0.494 0.006
Mean NSCCR amplitude 0.311 (0.372) 0.183 (0.295) 0.229 (0.215) 0.459 (0.637) 0.834 0.364 0.010
Lower- to higher-frequency EEG signal [(D+T)/(A+B)] 0.421 (0.200_ 0.374 (0.150) 0.290 (0.441) 0.079 (1.270) Fz REO 0.015 0.901 0.000
Resting-state frontal band power −1.970 (0.686) −1.300 (0.575) −1.771 (1.121) −1.862 (0.971) Fz REO 3.12 0.081 0.031
Left hemispheric lateralization of resting-state alpha activity 0.162 (0.281) 0.064 (0.360) −0.122 (0.510) 0.147 (0.495) F4-F3 REO 0.406 0.525 0.004
Open in a new tab

Manipulation check.

A mixed-factor GLM with period (normal breathing to paced HV) as the repeated measure and participant group (anxious/hypocapnic, anxious/normocapnic, heathy/hypocapnic, healthy/normocapnic) revealed a significant main effect of period, F1,98 = 43.34, p < .001, such that ETCO2 levels decreased following paced HV. There was no significant interaction, F3,98 = 13.37, p = .103, indicating that the degree of ETCO2 reduction did not vary by group.

Self-Report Measures.

On the HVQ-B cognitive subscale, there was no significant three-way interaction between period, anxiety, and hypocapnia. There was a significant two-way interaction between period and anxiety, F(1, 100) = 8.11, p = .005, η2p = .075, but not period and hypocapnia, F(1, 100) = 3.47, p = .065, η2p = .034. There was a significant simple main effect of period for anxious participants, F(1, 69) = 11.41, p = .001, such that scores on the HVQ-B cognitive subscale increased following paced HV. There was no simple main effect of period for healthy participants (p > .05). Results did not change substantively when controlling for SSRI/SNRI use.

On the HVQ-B affective subscale, there was a significant three-way interaction between period, anxiety, and hypocapnia, F(1, 100) = 4.39, p = .039, η2p = .042. There was a significant two-way interaction of period and anxiety among hypocapnic participants, F(1, 28) = 5.79, p = .023, but not among normocapnic participants, F(1, 72) = 5.63, p = .143. Examination of simple main effects revealed a significant increase following paced HV for anxious/hypocapnic participants, F(1, 19) = 12.28, p = .002, but not healthy/hypocapnic participants (p > .05). There was also a significant two-way interaction of period and hypocapnia among anxious participants, F(1, 68) = 12.02, p = .001, but not healthy participants, F(1, 32) = .672, p = .418. Examination of simple main effects revealed a significant increase following paced HV for anxious/hypocapnic participants as reported above, but not anxious/normocapnic participants (p > .05). Results did not change substantively when controlling for SSRI/SNRI use.

Thus, for anxious participants, paced HV results in an increase in subjective cognitive reactions to HV symptoms. For those anxious participants who are also hypocapnic at baseline, paced HV also increases affective responses to HV symptoms.

PNS Psychophysiological Measures.

For SCL, the period x anxiety x hypocapnia interaction was not significant. There was a significant two-way interaction between period and hypocapnia, F(1, 99) = 4.54, p = .036, η2p = .044, but not period and anxiety, F(1, 99) = .024, p = .876, η2p = .000. There was a significant simple main effect of period for hypocapnic participants, F(1, 29) = 4.30, p = .047, such that mean SCL levels increased following paced HV. There was no simple main effect of period for normocapnic participants, F(1, 72) = .413, p = 522. Results did not change substantively when controlling for SSRI/SNRI use.

There were no significant three-way or two-way interactions for NSSCR frequency (all ps > .05). For NSSCR amplitude, the period x anxiety x hypocapnia interaction was not significant and the period x anxiety and period x hypocapnia interactions were also not significant (all ps > .05); however, there was a significant anxiety x hypocapnia interaction. Anxious hypocapnic participants demonstrated significantly greater mean NSSCR amplitude collapsed across periods compared to healthy hypocapnic participants, F(1, 79) = 4.24, p = .043, and to anxious normocapnic participants, F(1, 79) = 6.59, p = .012. Results did not change substantively when controlling for SSRI/SNRI use.

Thus, for participants who are hypocapnic at baseline, paced HV results in increased sympathetic arousal as evidenced by SCL. As was the case in the Period 1 analysis, anxious participants who are hypocapnic at baseline show greater arousal as evidenced by NSSCR amplitude, though this was unaffected by a shift to paced HV.

EEG Measures.

For alpha asymmetry, although the three-way interaction (period x anxiety x hypocapnia) was not significant, the period x anxiety interaction approached significance (p = .056). There was a significant main effect of period for anxious participants, F(1, 66) = 4.207, p = .044, such that alpha asymmetry decreased following paced HV. The interaction no longer approached significance when controlling for SSRI/SNRI use, p = .300. There was no simple main effect of period on alpha asymmetry for healthy controls, F(1, 32) = 3.11, p = .087. In order to better understand these effects, we conducted the above analyses separately at electrodes F4 and F3. There was no significant three-way interaction between period, anxiety, and hypocapnia for Alpha 1 at F3 REO, F(1, 96) = .363, p = .548. There was a significant period x anxiety interaction, F(1, 96) = 4.092, p = .046, and a significant simple main effect of period for anxious participants, F(1, 66) = 6.240, p = .015, but not healthy participants, F(1, 32) = 1.738, p = .197, such that left-sided Alpha 1 at F3 REO increased following paced HV. There was no significant three-way interaction between period, anxiety, and hypocapnia for Alpha 1 at F4 REO, F(1, 96) = .000, p = .993. There were no significant two-way interactions (all ps > .05). The same analyses were conducted for REC and alpha 2 but no significant findings were observed.

There were no significant three-way (period x anxiety x hypocapnia) or two-way interactions (period x anxiety, period x hypocapnia) for ratio of lower- to higher-frequency EEG signal. However, when controlling for SSRI/SNRI use, a three-way interaction did emerge, F(1, 99) = 4.727, p = .032, η2p = .046. For the healthy group, there was no significant main effect of period (p = .129) and no period x hypocapnia interaction (p = .197). For the anxious group, there was a significant main effect of period (p = .009) and a significant period x hypocapnia interaction (p = .035). Follow-up tests indicated that among anxious participants, the ratio of lower- to higher-frequency EEG signal decreased significantly only for those who were normocapnic at baseline (p = .032).

There were no significant three-way (period x anxiety x hypocapnia) or two-way interactions (period x anxiety, period x hypocapnia) for resting-state frontal alpha band power (all ps > .05). Similar results were obtained when covarying for SSRI/SNRI use.

Thus, left-sided alpha frequency increases among anxious participants compared to healthy control participants following a period of paced HV; this shift was not dependent on hypocapnic status at baseline. The increase in left-sided alpha frequency is consistent with an increase in anxious arousal (Mathersul et al., 2008) and withdrawal motivation (Coan & Allen, 2003) following paced HV.

Discussion

Chronic HV (as evidenced by hypocapnia) is common in anxiety disorders, and appears to be associated with attenuated treatment response. The first aim of the present multi-method study was to examine the subjective, PNS, and CNS correlates of anxiety and baseline (chronic or trait) hypocapnia. By examining the psychological and biological markers of chronic hypocapnia, we hope to identify specific targets for future treatment development. We predicted that the specific combination of anxiety and baseline hypocapnia would be associated with a range of arousal-based measures. The second aim of the study was to determine whether these same biomarkers could be experimentally induced by paced HV, and whether anxious/hypocapnic participants would be particularly sensitive to this induction. By experimentally manipulating HV, it was hoped to establish causal relationships between HV and the biomarkers of arousal and to test individual differences in susceptibility to this induction.

The overall aims of the present study were to characterize the self-report, PNS, and CNS correlates of anxiety disorders, chronic hypocapnia, and their combination, and to examine how anxiety and chronic hypocapnia affect self-report, PNS, and CNS reactions to acute respiratory disturbance (paced HV). We predicted that anxious/hypocapnic participants would exhibit, at baseline, greater anxiety sensitivity (ASI-3) and hyperventilatory symptoms (NQ), greater PNS arousal (as measured by SCL and NSCCR), and several EEG indices of anxiety (lower-frequency signal, diminished alpha, and greater left alpha lateralization). We further predicted that when instructed to engage in paced HV, participants would report greater cognitive and affective reactions (HVQ-B), greater PNS arousal (SCL and NSSCR), and more lower-frequency signal, diminished alpha, and greater left alpha lateralization; and that these effects would be particularly pronouced for patients with anxiety disorders and chronic (baseline) hypocapnia, thus supporting a respiratory vulnerability model of anxiety.

Results were mixed, and our study hypotheses only partially supported. Under normal breathing conditions, we found that anxious patients, regardless of hypocapnic status, reported greater anxiety sensitivity and hyperventilatory symptoms. The finding of increased anxiety sensitivity in anxious patients is consistent with a substantial body of previous research (Olatunji & Wolitzky-Taylor, 2009). The fact that anxious patients’ report of HV symptoms was independent of chronic hypocapnia was unexpected, and suggests that the subjective experience of these symptoms may not be exclusively due to trait HV. This finding suggests a need for caution in interpreting self-report measures of physiological symptoms in the absence of psychophysiological recording.

We found evidence of PNS arousal as characterized by greater NSSCR amplitude among anxious/chronically hypocapnic participants compared to anxious/normocapnic participants. Thus, chronic hypocapnia adds to the objective physiological experience of anxiety beyond the effect of anxiety disorders per se. When combined with the heightened anxiety sensitivity described above, this finding may help explain the attenuated response to, and increased dropout from, cognitive-behavioral therapy (Davies & Craske, 2014; Tolin, Billingsley, et al., 2017). Specifically, chronically hypocapnic patients may experience greater sympathetic arousal than do normocapnic patients. Furthermore, the increased anxiety sensitivity among these patients may suggest that they have diminished capacity to tolerate this arousal, potentially leading to avoidant behavior. However, given that only one of the three hypothesized PNS variables differed between hypocapnic and normocapnic anxious participants, it will be important for future research to replicate and extend our findings to the investigation of treatment response and dropout.

Contrary to hypotheses, the present study did not find evidence of decreased resting-state frontal alpha band power (Wise et al., 2011) or greater left hemispheric lateralization of resting-state alpha band power (Nusslock et al., 2015), associated with either anxiety or with trait hypocapnia. We further found (only when controlling for medication use) a relatively higher EEG power in high-frequency bands in anxious patients compared to healthy controls. This was not consistent with our hypothesis of relatively greater lower-frequency power in chronically hypocapnic patients (Matteo et al., 1992). It could be argued, however, that relatively greater high-frequency power would be expected among anxious patients, given evidence that anxiety correlates positively with alpha and negatively with delta (Knyazev et al., 2004). Overall, the present findings raise questions about whether anxiety and chronic HV (as evidenced by baseline hypocapnia) are associated with robust alterations in EEG response. Further research is needed to examine the CNS correlates of anxiety, chronic HV, and their combination.

We then instructed participants to engage in paced HV. In examining the shift from baseline to paced HV, we found that paced HV leads to greater cognitive reactions among anxious patients, and greater affective response particularly among anxious/hypocapnic patients. These findings suggest, first, that anxiety (regardless of chronic hypocapnia) predisposes individuals to a greater cognitive reaction to respiratory disturbance. To the extent that anxious patients react to such disturbance with appraisals of being trapped or helpless, losing control, or passing out or having a heart attack (items from the HVQ-B), it is reasonable to surmise, based on cognitive-behavioral models of panic disorder, that this would result in increased anxious arousal, potentially predisposing these individuals to panic. We note that internal consistency of the HVQ-B cognitive scale was rather low, and therefore findings of increased cognitive reactions to HV among anxious participants should be interpreted cautiously. Second, the findings suggest that chronic HV (as evidenced by baseline hypocapnia), when combined with anxiety, presents a vulnerability to greater affective reactions (fear, anxiety, or nervousness, from the HVQ-B) to acute HV. We note again that anxious individuals who chronically hyperventilate have a high baseline level of anxiety sensitivity, and may therefore respond to respiratory disturbance with increased catastrophic cognitions and anxious arousal, which in turn may stimulate more acute HV and create a self-perpetuating cycle (see Margraf, 1993). As noted previously, we observed increased PNS arousal as indicated by mean SCL, but not NSSCR frequency or amplitude, among chronically hypocapnic participants due to acute HV. Unexpectedly, this effect did not depend on the presence of an anxiety disorder, despite reported increases in affective reactions to HV among anxious hypocapnic patients. While chronic HV may predispose hypocapnic individuals to experience increased PNS arousal following acute respiratory disturbance, it appears that only anxious participants are likely to report concurrent subjective emotional reactions.

Among anxious participants, paced HV increased left-sided alpha 1 power, thereby reversing the pattern of frontal asymmetry from R > L at baseline to L > R following HV. As noted in the Introduction, there is a robust literature suggesting that left lateralization is a marker of reduced approach-related motivation or increased withdrawal motivation (Nusslock et al., 2015), and is observed in community samples with anxious arousal (Mathersul et al., 2008) and in patients with panic disorder (Wise et al., 2011). Mathersul et al. (2008) and Harpfer et al. (2021) both reported that L > R lateralization was associated with anxious apprehension, whereas R > L lateralization was associated with anxious arousal. Similarly, Hofmann et al. (2005) found a relationship between L > R lateralization and worry induction (which presumably corresponds to anxious apprehension) in healthy volunteers, and Smith et al. (2016) found greater left alpha activation among undergraduate volunteers with elevated worry or with a generalized anxiety disorder diagnosis. Thus, the shift to left-sided lateralization may reflect an increased worry response (possibly worrying about physiological symptoms), rather than an arousal or panic response. This would further comport with the finding of increased cognitive reactions (e.g., worry) to HV as measured by the HVQ-B. Contrary to hypotheses, we did not find evidence of a greater ratio of lower- to higher-frequency EEG signal, as characterized by the parameter [(D+T)/(A+B)] (Matteo et al., 1992), as a function of paced HV or chronic hypocapnia. This contrasts with previous research demonstrating this effect in healthy volunteers (Kraaier et al., 1988), and additional replication is needed.

Conclusion

As noted previously, the study hypotheses were only partially supported, and several negative findings must be considered. Nevertheless, the positive findings from the present study provide preliminary support for a vulnerability model of anxiety disorders and chronic HV. Individuals with anxiety disorders exhibit greater anxiety sensitivity than do healthy controls, and they show greater baseline PNS arousal as measured by NSSCR amplitude (though not SCL or NSSCR frequency). We propose that this baseline anxiety sensitivity and physiological arousal predispose these individuals to experience greater cognitive and affective reactions to acute HV, as measured by the HVQ-B. Chronic hypocapnia did appear to result in greater PNS arousal (as evidenced by SCL, though not NSSCR) following acute HV, though this was true for healthy controls as well as anxious patients. It is not clear, therefore, whether this effect is specific to anxiety. We did find that anxious participants exhibited increased left-sided alpha power (though no change was noted in overall alpha power or lower- to higher-frequency EEG signal) following paced HV, though this was unrelated to baseline (chronic) HV. We would postulate that the baseline anxiety sensitivity seen in anxious patients makes them vulnerable to an anxious (possibly apprehension) reaction to acute HV, though clearly more work on this topic is needed.

The present results may help account for previous findings that chronic hypocapnia is associated with attenuated CBT response (Davies & Craske, 2014) and increased risk of treatment dropout (Tolin, Billingsley, et al., 2017). Patients who chronically hyperventilate appear to be more sensitive to respiratory distress, responding with higher levels of anxiety and poorer tolerance of the physiological sensations accompanying HV. It would be informative to determine whether addressing chronic HV (e.g., by using breathing retraining) can improve CBT outcomes and/or retention for hypocapnic patients. In light of findings that hyperventilation results in increased left frontal EEG alpha activity, we may cautiously posit that reduced efficacy of CBT in hypocapnic individuals could also result from disruption of information processing in response to arousal. Resting alpha activity is generally associated with neural inhibitory processes (Hartoyo et al., 2020), as first described by Hans Berger’s observation of a prominent decrease in alpha over the occipital region during transition from eye-closure to opening, suggesting an inverse association between scalp recorded alpha and regional cortical activity. Although a mechanistic interpretation cannot be drawn directly from current data, it is plausible to suggest that activity of frontal brain structures, supporting higher-order executive and working memory processes, is redirected as the brain contends with the acute effects of hypocapnia. Of note, frontal alpha asymmetry has been evaluated as a predictor and outcome of response to CBT in social anxiety, with higher left frontal activity at pretreatment associated with better outcome (Moscovitch et al., 2011).

Limitations

Several limitations of the present study should be noted. With enrollment attenuated by the COVID pandemic, the present study may not have been adequately powered to detect all effects. The study also consisted of many comparisons, raising the risk of cumulative Type 1 error. Although we were able to examine SSRI/SNRI use as a moderator, the relatively small sample precluded us from examining additional moderators. For example, we did not differentiate anxious participants by specific anxiety disorder, due to the low number of participants with each disorder. This particular issue may be relevant to findings concerning EEG alpha asymmetry, noting the findings of Mathersul and colleagues (2008) suggesting patterns of frontal EEG asymmetry may differ according to subtypes of anxiety, with right lateralization associated with anxious arousal, and left lateralization with anxious apprehension. Because HV has been more consistently linked in the literature with panic disorder, and because some of the prior literature on EEG responses have been limited to panic patients (Wise et al., 2011), it is possible that greater effects would have been seen in anxious patients with a primary diagnosis of panic disorder. One potential moderator for future study would be depression, which is associated with diminished right-sided frontal alpha (Bruder et al., 2017; Fingelkurts & Fingelkurts, 2015; Thibodeau et al., 2006); this appears to be the case in anxious samples with co-occurring depression as well (Bruder et al., 1997). Our 60-sec paced HV task was minimal, and greater subjective, PNS, and CNS effects might have been observed with a longer period of paced HV (Hornsveld et al., 1995).

Highlights.

  • The combination of anxiety and chronic hypocapnia is associated with greater baseline nonspecific skin conductance response amplitude

  • No baseline EEG abnormalities were noted as a function of anxiety or chronic hypocapnia

  • Anxiety and chronic hypocapnia are associated with negative cognitive appraisals of acute hyperventilation symptoms

  • Paced hyperventilation caused an increase in skin conductance levels among hypocapnic participants

  • Paced hyperventilation caused an increase in left-frontal EEG Alpha 1 power in anxious patients

Acknowledgments

Funded by NIMH grant R21 MH116187-01.

The authors thank Anishka Jean, Benjamin Katz, and Liya Mammo for assistance with data collection.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Bruder GE, Fong R, Tenke CE, Leite P, Towey JP, Stewart JE, McGrath PJ, & Quitkin FM (1997). Regional brain asymmetries in major depression with or without an anxiety disorder: a quantitative electroencephalographic study. Biological Psychiatry, 41(9), 939–948. 10.1016/S0006-3223(96)00260-0 [DOI] [PubMed] [Google Scholar]
  2. Bruder GE, Stewart JW, & McGrath PJ (2017). Right brain, left brain in depressive disorders: Clinical and theoretical implications of behavioral, electrophysiological and neuroimaging findings. Neuroscience and Biobehavioral Reviews, 78, 178–191. 10.1016/j.neubiorev.2017.04.021 [DOI] [PubMed] [Google Scholar]
  3. Carter MM, Sbrocco T, & Ayati F (2009). Predicting anxious response to a social challenge and hyperventilation: Comparison of the ASI and ASI-3. Journal of Behavior Therapy and Experimental Psychiatry, 40(3), 434–442. [DOI] [PubMed] [Google Scholar]
  4. Carter MM, Suchday S, & Gore KL (2001). The utility of the ASI factors in predicting response to voluntary hyperventilation among nonclinical participants. Journal of Anxiety Disorders, 15(3), 217–230. http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=11442140 [DOI] [PubMed] [Google Scholar]
  5. Coan JA, & Allen JJ (2003). Frontal EEG asymmetry and the behavioral activation and inhibition systems. Psychophysiology, 40(1), 106–114. http://www.ncbi.nlm.nih.gov/pubmed/12751808 [DOI] [PubMed] [Google Scholar]
  6. Davies CD, & Craske MG (2014). Low baseline pCO2 predicts poorer outcome from behavioral treatment: evidence from a mixed anxiety disorders sample. Psychiatry Research, 219(2), 311–315. 10.1016/j.psychres.2014.06.003 [DOI] [PubMed] [Google Scholar]
  7. Dawson ME, Schell AM, & Filion DL (2007). The electrodermal system. In Cacioppo JT, Tassinary LG, & Berntson GG (Eds.), Handbook of psychophysiology (3rd ed., pp. 159–181). Cambridge University Press. [Google Scholar]
  8. Dempsey JA, Forster HV, Gledhill N, & doPico GA (1975). Effects of moderate hypoxemia and hypocapnia on CSF [H+] and ventilation in man. Journal of Applied Physiology, 38(4), 665–674. http://www.ncbi.nlm.nih.gov/pubmed/237866 [DOI] [PubMed] [Google Scholar]
  9. Fingelkurts AA, & Fingelkurts AA (2015). Altered structure of dynamic electroencephalogram oscillatory pattern in major depression. Biological Psychiatry, 77(12), 1050–1060. 10.1016/j.biopsych.2014.12.011 [DOI] [PubMed] [Google Scholar]
  10. Gilbert C (1999). Hyperventilation and the body. Accident and Emergency Nursing, 7(3), 130–140. http://www.ncbi.nlm.nih.gov/pubmed/10693382 [DOI] [PubMed] [Google Scholar]
  11. Gonzalez-Bono E, Moya-Albiol L, Salvador A, Carrillo E, Ricarte J, & Gomez-Amor J (2002). Anticipatory autonomic response to a public speaking task in women: the role of trait anxiety. Biological Psychology, 60(1), 37–49. 10.1016/s0301-0511(02)00008-x [DOI] [PubMed] [Google Scholar]
  12. Guy W (1976). Assessment manual for psychopharmacology. U.S. Government Printing Office. [Google Scholar]
  13. Harpfer K, Spychalski D, Kathmann N, & Riesel A (2021). Diverging patterns of EEG alpha asymmetry in anxious apprehension and anxious arousal. Biological Psychology, 162, 108111. 10.1016/j.biopsycho.2021.108111 [DOI] [PubMed] [Google Scholar]
  14. Hartoyo A, Cadusch PJ, Liley DTJ, & Hicks DG (2020). Inferring a simple mechanism for alpha-blocking by fitting a neural population model to EEG spectra. PLoS Comput Biol, 16(4), e1007662. 10.1371/journal.pcbi.1007662 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hegel MT, & Ferguson RJ (1997). Psychophysiological assessment of respiratory function in panic disorder: evidence for a hyperventilation subtype. Psychosomatic Medicine, 59(3), 224–230. http://www.ncbi.nlm.nih.gov/pubmed/9178332 [DOI] [PubMed] [Google Scholar]
  16. Hofmann SG, Moscovitch DA, Litz BT, Kim HJ, Davis LL, & Pizzagalli DA (2005). The worried mind: autonomic and prefrontal activation during worrying. Emotion, 5(4), 464–475. 10.1037/1528-3542.5.4.4643542.5.4.464 [DOI] [PubMed] [Google Scholar]
  17. Hofmann SG, & Smits JA (2008). Cognitive-behavioral therapy for adult anxiety disorders: a meta-analysis of randomized placebo-controlled trials. Journal of Clinical Psychiatry, 69(4), 621–632. 10.4088/jcp.v69n0415 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Hornsveld H, Garssen B, & van Spiegel P (1995). Voluntary hyperventilation: the influence of duration and depth on the development of symptoms. Biological Psychology, 40(3), 299–312. http://www.ncbi.nlm.nih.gov/pubmed/7669838 [DOI] [PubMed] [Google Scholar]
  19. Klein DF (1993). False suffocation alarms, spontaneous panics, and related conditions. An integrative hypothesis. Archives of General Psychiatry, 50(4), 306–317. http://www.ncbi.nlm.nih.gov/pubmed/8466392 [DOI] [PubMed] [Google Scholar]
  20. Klimesch W (1999). EEG alpha and theta oscillations reflect cognitive and memory performance: a review and analysis. Brain Research. Brain Research Reviews, 29(2–3), 169–195. http://www.ncbi.nlm.nih.gov/pubmed/10209231 [DOI] [PubMed] [Google Scholar]
  21. Knyazev GG, Savostyanov AN, & Levin EA (2004). Alpha oscillations as a correlate of trait anxiety. International Journal of Psychophysiology, 53(2), 147–160. 10.1016/j.ijpsycho.2004.03.001 [DOI] [PubMed] [Google Scholar]
  22. Kraaier V, van Huffelen AC, & Wieneke GH (1988). Changes in quantitative EEG and blood flow velocity due to standardized hyperventilation; a model of transient ischaemia in young human subjects. Electroencephalography and Clinical Neurophysiology, 70(5), 377–387. http://www.ncbi.nlm.nih.gov/pubmed/2460311 [DOI] [PubMed] [Google Scholar]
  23. Kreibig SD, Wilhelm FH, Roth WT, & Gross JJ (2007). Cardiovascular, electrodermal, and respiratory response patterns to fear- and sadness-inducing films. Psychophysiology, 44(5), 787–806. 10.1111/j.1469-8986.2007.00550.x [DOI] [PubMed] [Google Scholar]
  24. Ley R (1985). Blood, breath, and fears: A hyperventilation theory of panic attacks and agoraphobia. Clinical Psychology Review, 5(4), 271–285. [Google Scholar]
  25. Lykken DT, & Venables PH (1971). Direct measurement of skin conductance: a proposal for standardization. Psychophysiology, 8(5), 656–672. http://www.ncbi.nlm.nih.gov/pubmed/5116830 [DOI] [PubMed] [Google Scholar]
  26. Margraf J (1993). Hyperventilation and panic disorder: A psychophysiological connection. Advances in Behaviour Research and Therapy, 15, 49–74. [Google Scholar]
  27. Mathersul D, Williams LM, Hopkinson PJ, & Kemp AH (2008). Investigating models of affect: relationships among EEG alpha asymmetry, depression, and anxiety. Emotion, 8(4), 560–572. 10.1037/a0012811 [DOI] [PubMed] [Google Scholar]
  28. Matteo RS, Ornstein E, Schwartz AE, Young WL, Weinstein J, & Cain CF (1992). Effects of hypocarbia on the pharmacodynamics of sufentanil in humans. Anesthesia and Analgesia, 75(2), 186–192. http://www.ncbi.nlm.nih.gov/pubmed/1385931 [DOI] [PubMed] [Google Scholar]
  29. Meuret AE, Rosenfield D, Wilhelm FH, Zhou E, Conrad A, Ritz T, & Roth WT (2011). Do unexpected panic attacks occur spontaneously? Biological Psychiatry, 70(10), 985–991. 10.1016/j.biopsych.2011.05.027 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Moscovitch DA, Santesso DL, Miskovic V, McCabe RE, Antony MM, & Schmidt LA (2011). Frontal EEG asymmetry and symptom response to cognitive behavioral therapy in patients with social anxiety disorder. Biological Psychology, 87(3), 379–385. 10.1016/j.biopsycho.2011.04.009 [DOI] [PubMed] [Google Scholar]
  31. Munjack DJ, Brown RA, & McDowell DE (1993). Existence of hyperventilation in panic disorder with and without agoraphobia, GAD, and normals: Implications for the cognitive theory of panic. Journal of Anxiety Disorders, 7(1), 37–48. [Google Scholar]
  32. Muthuraju S, & Pati S (2014). Effect of hypobaric hypoxia on cognitive functions and potential therapeutic agents. Malaysian Journal of Medical Science, 21(Spec Issue), 41–45. https://www.ncbi.nlm.nih.gov/pubmed/25941462 [PMC free article] [PubMed] [Google Scholar]
  33. Nikula R (1991). Psychological correlates of nonspecific skin conductance responses. Psychophysiology, 28(1), 86–90. 10.1111/j.1469-8986.1991.tb03392.x [DOI] [PubMed] [Google Scholar]
  34. Nusslock R, Walden K, & Harmon-Jones E (2015). Asymmetrical frontal cortical activity associated with differential risk for mood and anxiety disorder symptoms: An RDoC perspective. International Journal of Psychophysiology, 98(2 Pt 2), 249–261. 10.1016/j.ijpsycho.2015.06.004 [DOI] [PubMed] [Google Scholar]
  35. Olatunji BO, & Wolitzky-Taylor KB (2009). Anxiety sensitivity and the anxiety disorders: a meta-analytic review and synthesis. Psychological Bulletin, 135(6), 974–999. 10.1037/a0017428 [DOI] [PubMed] [Google Scholar]
  36. Rapee R (1986). Differential response to hyperventilation in panic disorder and generalized anxiety disorder. Journal of Abnormal Psychology, 95(1), 24–28. http://www.ncbi.nlm.nih.gov/pubmed/3084604 [DOI] [PubMed] [Google Scholar]
  37. Rosenfield D, Zhou E, Wilhelm FH, Conrad A, Roth WT, & Meuret AE (2010). Change point analysis for longitudinal physiological data: detection of cardio-respiratory changes preceding panic attacks. Biological Psychology, 84(1), 112–120. 10.1016/j.biopsycho.2010.01.020 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Sabourin BC, Stewart SH, Watt MC, & Macdonald AB (2013). Development and psychometric evaluation of a brief version of the hyperventilation questionnaire: the HVQ-B. Cognitive Behaviour Therapy, 42(3), 193–202. 10.1080/16506073.2012.756059 [DOI] [PubMed] [Google Scholar]
  39. Sato W, Kochiyama T, & Yoshikawa S (2020). Physiological correlates of subjective emotional valence and arousal dynamics while viewing films. Biological Psychology, 157, 107974. 10.1016/j.biopsycho.2020.107974 [DOI] [PubMed] [Google Scholar]
  40. Schubert A, & Drummond JC (1986). The effect of acute hypocapnia on human median nerve somatosensory evoked responses. Anesthesia and Analgesia, 65(3), 240–244. http://www.ncbi.nlm.nih.gov/pubmed/3082246 [PubMed] [Google Scholar]
  41. Sharma S, & Hashmi MF (2021). Hypocarbia. In StatPearls. https://www.ncbi.nlm.nih.gov/pubmed/29630219 [PubMed] [Google Scholar]
  42. Smith EE, Zambrano-Vazquez L, & Allen JJ (2016). Patterns of alpha asymmetry in those with elevated worry, trait anxiety, and obsessive-compulsive symptoms: A test of the worry and avoidance models of alpha asymmetry. Neuropsychologia, 85, 118–126. 10.1016/j.neuropsychologia.2016.03.010 [DOI] [PubMed] [Google Scholar]
  43. Smoller JW, Pollack MH, Otto MW, Rosenbaum JF, & Kradin RL (1996). Panic anxiety, dyspnea, and respiratory disease. Theoretical and clinical considerations. American Journal of Respiratory and Critical Care Medicine, 154(1), 6–17. 10.1164/ajrccm.154.1.8680700 [DOI] [PubMed] [Google Scholar]
  44. Taylor S, Zvolensky MJ, Cox BJ, Deacon B, Heimberg RG, Ledley DR, Abramowitz JS, Holaway RM, Sandin B, Stewart SH, Coles M, Eng W, Daly ES, Arrindell WA, Bouvard M, & Cardenas SJ (2007). Robust dimensions of anxiety sensitivity: development and initial validation of the Anxiety Sensitivity Index-3. Psychological Assessment, 19(2), 176–188. 10.1037/1040-3590.19.2.176 [DOI] [PubMed] [Google Scholar]
  45. Thibodeau R, Jorgensen RS, & Kim S (2006). Depression, anxiety, and resting frontal EEG asymmetry: a meta-analytic review. Journal of Abnormal Psychology, 115(4), 715–729. 10.1037/0021-843X.115.4.715 [DOI] [PubMed] [Google Scholar]
  46. Tolin DF, Billingsley AL, Hallion LS, & Diefenbach GJ (2017). Low pre-treatment end-tidal CO2 predicts dropout from cognitive-behavioral therapy for anxiety and related disorders. Behaviour Research and Therapy, 90, 32–40. 10.1016/j.brat.2016.12.005 [DOI] [PubMed] [Google Scholar]
  47. Tolin DF, Gilliam C, Wootton BM, Bowe W, Bragdon LB, Davis E, Hannan SE, Steinman SA, Worden B, & Hallion LS (2018). Psychometric properties of a structured diagnostic interview for DSM-5 anxiety, mood, and obsessive-compulsive and related disorders. Assessment, 25(1), 3–13. 10.1177/1073191116638410 [DOI] [PubMed] [Google Scholar]
  48. Tolin DF, McGrath PJ, Hale LR, Weiner DN, & Gueorguieva R (2017). A multisite benchmarking trial of capnometry assisted biofeedback training for panic disorder in naturalistic treatment settings. Applied Psychophysiology and Biofeedback, 42(1), 51–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. van den Hout MA, Hoekstra R, Arntz A, Christiaanse M, Ranschaert W, & Schouten E (1992). Hyperventilation is not diagnostically specific to panic patients. Psychosomatic Medicine, 54(2), 182–191. http://www.ncbi.nlm.nih.gov/pubmed/1565755 [DOI] [PubMed] [Google Scholar]
  50. van Dixhoorn J, & Duivenvoorden HJ (1985). Efficacy of Nijmegen Questionnaire in recognition of the hyperventilation syndrome. Journal of Psychosomatic Research, 29(2), 199–206. http://www.ncbi.nlm.nih.gov/pubmed/4009520 [DOI] [PubMed] [Google Scholar]
  51. van Dixhoorn J, & Folgering H (2015). The Nijmegen Questionnaire and dysfunctional breathing. ERJ Open Res, 1(1). 10.1183/23120541.00001-2015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Wilhelm FH, Trabert W, & Roth WT (2001). Physiologic instability in panic disorder and generalized anxiety disorder. Biological Psychiatry, 49(7), 596–605. http://www.ncbi.nlm.nih.gov/pubmed/11297717 [DOI] [PubMed] [Google Scholar]
  53. Wise V, McFarlane AC, Clark CR, & Battersby M (2011). An integrative assessment of brain and body function ‘at rest’ in panic disorder: a combined quantitative EEG/autonomic function study. International Journal of Psychophysiology, 79(2), 155–165. 10.1016/j.ijpsycho.2010.10.002 [DOI] [PubMed] [Google Scholar]

RESOURCES