Anxiety and autonomic flexibility: a cardiovascular approach1

Biological Psychology 47 (1998) 243 – 263

Anxiety and autonomic flexibility: a cardiovascular approach1 Bruce H. Friedman *, Julian F. Thayer Department of Psychology, The Pennsyl6ania State Uni6ersity, PA, USA

Abstract Autonomic characteristics of panickers, blood phobics, and nonanxious controls were compared with a variety of cardiovascular measures, including spectral analysis of the cardiac inter-beat interval time series (derived from the electrocardiogram). Responses to laboratory stressors (shock avoidance and cold face stress) of 16 participants who reported recent occurrences of frequent severe panic attacks, 15 participants who reported strong somatic reactions and fainting to the sight of blood, and 15 controls, were recorded. Results suggested distinct autonomic patterns among the three groups. Across conditions, panickers displayed the highest heart rates (HR) coupled with the least HR variability, which indicates low levels of cardiac vagal tone. Blood phobics showed more vagally mediated HR variability than panickers, with a significant association between cardiac rate and mean arterial pressure. Controls generally showed the most HR variability and ‘spectral reserve’ (a quality that indicates flexible responsivity). Results are discussed in the context of traditional models of anxiety and autonomic activity in contrast to contemporary notions of stability and change in biological systems. © 1998 Elsevier Science B.V. All rights reserved. Keywords: Anxiety; Panic attacks; Autonomic activity; Cardiovascular study

* Corresponding author. Present address: Department of Psychology, Virginia Polytechnic Institute and State University, Blacksburg, VA 24060, USA. E-mail: [email protected] 1 Portions of this paper were presented in J.F. Thayer (Chair), New Approaches to Cardio6ascular Reacti6ity Symposium conducted at the 33rd Annual Meeting of the Society for Psychophysiological Research, October 1993, Rottach-Egern, Germany. This study was conducted in partial fulfillment of the requirements of the doctoral dissertation of the first author. 0301-0511/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved. PII S 0 3 0 1 - 0 5 1 1 ( 9 7 ) 0 0 0 2 7 - 6

244

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

1. Anxiety and autonomic dysregulation Anxiety is often accompanied by somatic manifestations that suggest marked changes in autonomic nervous system (ANS) activity, such as rapid heart rate (HR), shortness of breath, and sweating. These symptoms have frequently been viewed as signs of increased sympathetic (SNS) activation. This interpretation is in accord with Cannon’s ‘flight-or-fight’ model of fear (Cannon, 1929) that is based upon the assumption of global states of SNS activation. Among the various forms of anxiety, panic is singularly marked by the rapid onset of intense somatic symptoms and subjective reports of terror (Barlow, 1988). The presence of recurrent uncued panic attacks is the primary defining feature of panic disorder (PD) (American Psychiatric Association, 1994). Descriptions of this syndrome have an extended history in the medical literature and were often subsumed under the generic label of ‘neurocirculatory asthenia’ (NCA, see Caranasos, 1974, for review). This diagnosis was given to patients who experienced bouts of severe anxiety accompanied by pronounced somatic disturbances without apparent organic cause. The most frequently reported symptom in these cases was tachycardia. Similarities in symptomology suggest a substantial diagnostic overlap between NCA and PD (Stampler, 1982). The scope of biological models of panic has been exceedingly broad. However, the salience of reports of tachycardia have implicated SNS disturbances since early investigations of panic syndromes (e.g. Fraser and Wilson, 1918). More contemporary research has been focused at the level of adrenergic neurotransmitter dysfunctions, but has not yielded definitive results (e.g. Grunhaus et al., 1981). More global characterizations have invoked the concept of an autonomic ‘imbalance’ in the direction of sympathetic dominance in pathological anxiety (e.g. Reich, 1982 (original work published in 1939)). Other broad ANS representations include slow habituation (Lader, 1980), high tonic activation (Roth et al., 1986), and other sundry ANS dysfunctions (for reviews of ANS models of panic, see Mitchell and Shapiro, 1991; Friedman and Thayer, in press). Of particular bearing on the present study are models which have linked anxiety with excess autonomic lability and reactivity (Eysenck, 1970; Costello, 1971). In such depictions, the ANS is seen as hyperreactive with chronic fluctuations from a steady state. These representations are in need of reconsideration in view of the present state of knowledge of the specificity of ANS regulation of cardiovascular (CV) function, as is detailed below.

1.1. Anxiety and autonomic specificity There are several overarching concerns with the research on the autonomic underpinnings of panic anxiety. One issue is the prevailing emphasis on the SNS and the relative neglect of parasympathetic (PNS) activity. Recent advances in both cardiology and psychophysiology have stressed the importance of vagal influences on CV regulation (e.g. Saul, 1990; Porges, 1995). Furthermore, studies of adrenergic

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

245

dysfunction have typically involved the use of pharmacological agents that may alter significant ANS interactions in CV control (cf. Levy, 1990). Consequently, research that has been focused exclusively on sympathetic aberrations is of limited scope due to the omission of the role of the PNS. Another obstacle is the widespread practice of depicting ANS activity in global terms. Vague depictions like ‘autonomic hyperreactivity’ persist in spite of abundant evidence that ANS activity is expressed in exquisitely complex patterns (Lacey, 1967; Berntson et al., 1991; Wolf, 1995). With particular reference to the cardiac symptomatology of panic, sympathetic and vagal activity are known to differentially affect the dimensions of rate, contractility, and conduction velocity (Levy, 1990; Huang et al., 1995). Thus, models that are based on generalized autonomic activity oversimplify and misrepresent the complexity of ANS functioning. The prominence of tachycardia in panic attacks raises the spectre of an irregularity in autonomic control of HR. This suspicion arises from the primary role that ANS activity plays in HR regulation (Berne and Levy, 1992). Vagal control is more efficient than sympathetic in modulating rapid changes in HR due to the superior frequency response characteristics of cholinergic over adrenergic neurotransmitter systems (Saul, 1990). A high degree of vagal control allows for enhanced responsivity and sensitivity of HR to changing environmental demands (Goldberger, 1991). Accordingly, increased cardiac vagal tone has generally been found to be associated with elevated HR variability (Stein et al., 1994). High vagal tone may also provide protection against malignant cardiac arrhythmias (Saini and Verrier, 1989; Kamarck and Jennings, 1991). Such findings suggest that a deficiency in vagal HR regulation may be more important than a sympathetic dysfunction in susceptibility to the tachycardia of panic attacks. This speculation was made over 50 years ago with little impact on subsequent research (Wood, 1941). However, contemporary research has generally found both diminished HR variability and reduced vagal tone to be associated with panic (see Friedman and Thayer, in press, for review). Furthermore, provocative evidence can be found that vagal enhancement techniques can ameliorate panic (Sartory and Olajide, 1988), and that positive therapy outcome for anxiety is associated with increased cardiac vagal tone and HR variability (Friedman et al., 1993a; Middleton and Ashby, 1995). The majority of studies of autonomic dysfunction in panic have relied on inferences drawn from pharmacological manipulations rather than the non-invasive assessment of cardiac regulatory mechanisms. This latter area has stimulated a great deal of recent interest in both cardiology and psychophysiology (for reviews, see Stein et al., 1994; van Ravenswaaij-Arts et al., 1993; Jennings and McKnight, 1994; Porges, 1995). Investigation of the relative SNS and PNS components of CV control with these methods may be helpful in clarifying the ANS underpinnings of panic.

246

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

2. Anxiety, reactivity, and organismic self-regulation

2.1. Homeostasis and homeodynamics Models of anxiety that are based on excess autonomic lability and ANS hyperreactivity are grounded in a conventional homeostatic view of organismic functioning (Cannon, 1939). That is, these perspectives hold that the ANS in chronic anxiety is overreactive with frequent perturbations from a steady state. This notion seems plausible due to the rapid onset of symptoms that are the hallmark of panic attacks; i.e. such incidents logically suggest disturbances of steady-state functioning. However, alternative views of organismic self-regulation have challenged the notion that physiological variables strive for rigid constancy. For example, stability in one physiological system can be supported by variability in another (heterostasis; Davis, 1958). A range of changing values may be also maintained, rather than a single set point (homeorhesis; Stallone and Stunkard, 1991). In application to CV activity, fluctuating processes that maintain organismic stability have been termed homeodynamic (Appel et al., 1989). In the homeodynamic framework, CV integrity is marked by complex temporal variability, and extreme predictability of HR is associated with pathology (Pincus et al., 1991; Latson, 1994; Peng et al., 1994). Short-term fluctuations in HR are due to feedback from a variety of internal mechanisms such as respiration, blood pressure (BP) control, thermoregulation, blood volume modulation, and the renin–angiotensin system (Saul, 1990; Goldberger, 1991). These rhythms are primarily mediated by the ongoing interplay of cardiac sympathetic and vagal activity. Spectral analysis of the electrocardiogram (ECG) and analogous methods of analysis of cardiac inter-beat interval (IBI) sequences have been used to distinguish among the intrinsic sources of variability in the HR time series (for reviews, see Appel et al., 1989; Malik and Camm, 1990; Malliani et al., 1990; Stein et al., 1994). These fluctuations occur at different frequencies, which allows for differential mapping of the components of the power spectra onto autonomic indices of HR mediation. A relatively fast, high-frequency (HF) component ( 0.25 Hz) which corresponds to the frequency of respiration, and a slower, low-frequency (LF) component (0.10 Hz) which primarily reflects baroreceptor-mediated regulation of BP, have consistently been found in the power spectra of the ECG and the IBI time series. Except at very slow breathing frequencies, the respiratory component is exclusively mediated by vagal (cholinergic) activity, which is broad-band and rapid in its effects (Saul, 1990). The BP component reflects substantial amounts of sympathetic control with varying levels of vagal influence; the relative contributions of each are a current point of controversy (for reviews, see Pagani et al., 1992; Friedman and Thayer, in press). HR response to adrenergic stimulation is narrow band, with negligible impact at frequencies above 0.10 Hz and a delay in effect. Thus, sympathetic activity has much less influence in regulating short-term changes in HR (Saul, 1990).

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

247

2.2. Flexibility, 6agal tone, and spectral reser6e Spectral analysis of the ECG and IBI time series from healthy subjects typically generates a wide power frequency distribution that has been labeled the ‘spectral reserve’ of the system (Goldberger and Rigney, 1990; West, 1990). This broad spectrum has a 1/f-like distribution (inverse power law) and is indicative of multiple sources of variance (Goldberger et al., 1985). In contrast, pathological states show a loss in overall power, a narrowing of the frequency spectra to a few sharp peaks, and a relative decrease in HF power (Goldberger and West, 1987). An estimate of spectral reserve can be obtained by plotting these spectra on double-log axes of power and frequency, and calculating the slope of the regression line fitted to these points (Lipsitz et al., 1990). Steeper slopes correspond to narrow-band spectra, diminished HF power, and less spectral reserve (Goldberger, 1990)2. These attributes mark a system that lacks flexibility and adaptability. Hence, rigid regularity in the HR time series indicates a loss of the sources of information that normally increase complex variability in the HR signal. Such ‘pathological periodicity’ has been found in the spectra of a variety of physiological measures across a wide range of pathologies, and may also generally characterize the aging process (Mandell and Schlesinger, 1990; Kaplan et al., 1991). Furthermore, diminished spectral reserve is also found in laboratory conditions that elicit sympathetic activation and vagal withdrawal (Friedman et al., 1996). These findings are consistent with developmental research that has found cardiac vagal tone to be associated with adaptive responsivity to the environment. Infants and young children with high vagal tone are superior in the ability to shift and sustain attention (Porges, 1992). Neonates low in HR variability appear to be generally less responsive to a variety of stimuli, and lack the facility to modulate emotion (Porges, 1991). Extremely shy children who may be at risk to develop adult anxiety disorders have also been shown to consistently display high and stable HRs (Kagan et al., 1990). In sum, high vagal tone seems to be a marker for physiological and psychological flexibility. Alternatively, reductions in the complexity of responding in a wide range of physiological channels are associated with poor health outcomes and a lack of adaptive variability in behavioral and cognitive functioning. These process parallels across levels of analysis are consistent with the principle of self-similarity which describes correspondences in structure and process across multiple scales in nature (West, 1990). Viewed from this perspective, panic may be associated with decreased, rather than increased autonomic lability, particularly as indicated by HR variability indices. PD (especially when associated with agoraphobia) represents a loss of options; it may be regarded as a fear-related, stereotypic cognitive/behavioral/somatic complex that inhibits exploration of the environment. An early formulation 2

Though it has been suggested that this measure may, under certain conditions, have a relationship to the correlation dimensions (Deering and West, 1992), such conditions are rarely met in typical psychophysiological experiments (Thayer and Friedman, 1993).

248

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

of this hypothesis portrayed anxiety as a state of general organismic withdrawal that is characterized by sympathetic dominance (Reich, 1982 (original work published in 1939)). It is often held that electrodermal (EDA) lability is a correlate of anxiety (e.g. Cruz and Larsen, 1995). This notion is supported by oft-cited research that has found greater spontaneous fluctuations in and diminished habituation of EDA in anxiety states (for review, see Lader, 1980). However, there have been some failures to replicate this relationship (for review, see Katkin, 1975), and even a negative relationship between EDA lability and state anxiety has been reported (Kelsey, 1991). In the latter study, EDA labiles also showed more rapid habituation of cardiac reactivity to stress, which suggests adaptive environmental responsiveness. Moreover, EDA lability has been found to be associated with superior performance on perceptual vigilance tasks, efficient allocation of attentional resources, and enhanced responsivity to changing environmental demands (Surwillo and Quilter, 1965; Katkin, 1975; Sostek, 1978; Schell et al., 1988). These qualities are not typically associated with anxiety, which on the contrary, has been linked with attentional deficits (Mathews, 1990) and poor adaptive responsiveness to the environment (Neary and Zuckerman, 1976). Indeed, a diminished range of EDA responding has been found in both panic and generalized anxiety disorders (Hoehn-Saric et al., 1989, 1991), and were interpreted as evidence of reduced autonomic flexibility in both groups. This narrowed range is consistent with reports of decreased EDA orienting responses in high state anxious individuals (Neary and Zuckerman, 1976), and lower EDA stress reactivity in clinically anxious women (Goldstein, 1965). In sum, the notion of excess autonomic lability in anxiety appears to be generally belied by EDA as well as CV data. It may be that diminished variability in peripheral physiological responding is common to various types of clinical anxiety. Reduced HR variability and vagal tone have been reported in generalized anxiety disorder and worry (Lyonfields et al., 1995; Thayer et al., 1996a). In addition to a restricted range of EDA, a reduced range in HR and electromyographic responding have been found in GAD, and have been interpreted as reflecting a generalized psychophysiological response rigidity to environmental stimuli (Hoehn-Saric et al., 1989; Hazlett et al., 1994). Collectively, these findings led to the present hypothesis that the tachycardia of panic may be related to poor autonomic HR control as reflected in diminished HR variability, HF power, and spectral reserve, in contrast to portrayals of excess autonomic lability in anxiety.

2.3. Panic and blood phobia A striking dissimilarity exists between the CV symptoms observed in two types of anxiety disorders: PD and blood–injection–injury phobia. A panic attack is manifested as an extreme ‘hyperarousal’ of the organism, while the phobic response to blood stimuli often ends in a dramatic deacti6ation of the organism—fainting (vasovagal syncope). Vasovagal syncope is a diphasic response; a brief period of hypertension and tachycardia is followed by bradycardia and hypotension, which can result in an inadequate cerebral blood supply and a loss of consciousness (Ost

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

249

et al., 1984; Thyer and Curtis, 1985). The autonomic underpinnings of this reaction are presumed to be sympathetic hyperarousal immediately followed by sympathetic inhibition and concomitant overcompensatory parasympathetic rebound (Glick and Yu, 1963; Carruthers and Taggart, 1973). This pattern is highly dissimilar to a panic attack, which gives evidence of elevated sympathetic activity and vagal withdrawal. Thus, panickers and blood phobics may be distinguished by their CV control mechanisms. This speculation was supported in several recent studies that found relative sympathetic dominance in panickers and more vagal influence in blood phobics, as revealed by HR variability indices (Friedman et al., 1993b; Tyrrell et al., 1995). The present study was conducted to replicate these findings and to extend them to a non-anxious control group. Tasks were selected to induce HR variability across a wide range of ANS activity. The data were aggregated across tasks and within persons in line with recommendations that response patterns in individuals are best gauged with multiple appraisals over different situations (Rosenzweig, 1956; Epstein and O’Brien, 1985). Specifically, it was predicted that panickers would show the most sympathetic and least vagal control of HR as indicated by reductions in: (1) HR variability; (2) HF spectral power relative to LF; and (3) spectral reserve.

3. Method

3.1. Participants An analogue sample of clinical panickers and blood phobics was recruited by questionnaire for the study. Panic attacks have been found to occur with a relatively high frequency in the general population, and non-clinical panickers appear to closely resemble PD patients on symptom criteria (Norton et al., 1986). Furthermore, the validity of analogue clinical samples in experimental studies of this nature is well supported (Borkovec and Rachman, 1979). Finally, a major goal of the present study was to replicate Friedman et al. (1993a) and Tyrrell et al. (1995) in which non-clinical samples were used, so the selection procedures for the anxiety groups were identical to those reported in those studies. Forty-six students were recruited from an undergraduate psychology course to participate in the study (16 panickers (11 female and five male); 15 blood phobics (12 female and three male); 15 controls (seven female and eight male); mean age 19.3 years)3. Participants were selected on the basis of two questionnaires: one 3 In order to approximate the clinical disorders as closely as possible, participants with the most extreme questionnaire scores were selected without regard to gender. The gender ratios obtained in the present samples likely reflect the higher rates for females of both PD and blood phobia that have been found in the general population (Ost et al., 1984; Barlow, 1988). Indeed, nonclinical panic has been found to more closely resemble its clinical analogue in females (Wilson et al., 1992). The sole criteria for the control group was a lack of both panic and blood phobia; as such, a more equal gender ratio was obtained in this sample. However, (1) there was no a priori reason to suspect gender difference on the CV variables used in this paradigm that involves data aggregation across diverse conditions, and (2) deletion of the male data from the groups did not substantively alter the results.

250

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

relevant to panic attacks and the other to blood phobia. Those students who scored the highest on the respective questionnaires were chosen for the panic and blood phobic sample from the total pool of 1500 questionnaires. The panic questionnaire was constructed to mirror the DSM-III-R4 (American Psychiatric Association, 1987) criteria for PD on self-reported frequency and intensity of panic attacks as closely as possible, based on a instrument previously developed for this same purpose (Michelson et al., 1988). All participants reported at least one spontaneous attack of intense fear within the last month, and between one and nine attacks in the last 2 months (mean number of attacks in the last 2 months=3.2, S.D.=2.1). Thus, the DSM-III-R criterion of at least four attacks in the last month for a diagnosis of PD was closely approximated. The intensity of the 13 DSM-III-R symptoms of panic were rated on an eight-point Likert-type scale, with 0=symptom not present, 4= moderate intensity, and 8=very severe intensity. The ratings were summed and a composite score calculated for each subject (mean score= 32.4, S.D.= 15.8). The obtained ratings were similar to those obtained by Friedman et al. (1993b) and Tyrrell et al. (1995)5. On the blood phobia questionnaire, participants rated the intensity of nausea, faintness, sweating, and subjective feeling of anxiety at the sight of blood on a similar eight-point Likert-type scale. These scores were summed and a composite score was calculated (mean score=18.2, S.D.= 5.6). Importantly, all participants in this group reported prior incidents of fainting to the sight of blood. The control group was defined as follows: no reported panic attacks within the last 2 months, no incidents of fainting to the sight of blood, and a blood phobia composite score of less than 8.0 (a very small amount of disturbance in response to blood stimuli). All participants reported no health problems and had abstained from caffeine, nicotine, and alcohol for at least 12 h prior to the experiment. Informed consent was obtained prior to, and course credit was given for, participation in the study.

3.2. Apparatus Ag – AgCl electrodes (UFI Corp.) were used to record the ECG (electrode impedances reduced to less than 10 kV with Omniprep skin prep; Omniprep, Inc.). The ECG was amplified with Grass 7P3 preamplifiers and Grass 7DA driver amplifiers (Grass Instruments, Inc.). BP was monitored with a SD-700A BP/Pulse Monitor (IBS Corp.). Spectral analysis of the IBIs (derived from the ECG signal;

4

This study was conducted prior to publication of DSM-IV. Panic Disorder Questionnaire: mean = 39.1, S.D. =9.5; Blood Phobia Questionnaire: mean = 20.9, S.D.= 4.8. 5

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

251

see below for details) was performed on a Swan 386/20D PC with software that employed an autoregressive algorithm (Colombo et al., 1990). Stimuli were generated on an Apple IIE computer and displayed on an Apple monitor. A Mark I Behavior Modifier (Farrall Instruments), which produces voltages that can be varied in 10 steps of intensity from 0.123 to 0.657 VAC, was used in the shock-avoidance task.

3.3. Procedure The procedures for this experiment were identical to those reported in Friedman et al. (1993a,b), Tyrrell et al. (1995) (Experiment 1), and Muth et al. (1996), and will not be repeated here in detail. Briefly, the experiment involved a stimulation paradigm in which each participant was exposed to manipulations that were selected to perturb ongoing system functioning (see Levy and Zieske, 1969; Uijtdehaage and Thayer, 1989). The dynamic response to and recovery from these perturbations formed the basic data set. This procedure differs from the typical static experimental design in which the level of response during each discrete condition is the object of study. The former method, in conjunction with the pooled cross-sectional time series approach to the data analysis, allowed for the assessment of group differences in dynamic autonomic control by the estimation of the coupling parameters (within-group across-person correlations) among the CV variables. The first experimental period was simple relaxation. The order of the next two tasks (reaction time/shock avoidance (RT) and cold face stress (CF)) was counterbalanced within groups and across individuals. The last task was always combined RT and CF (RT/CF). Each task lasted for 4 min, and was immediately followed by a 4-min recovery period of simple relaxation. Only one shock (which followed the first response in the third minute) was actually administered during each RT task (RT and RT/CF) in order to maintain salience of the threat of shock. BP was taken at 1-min intervals during all tasks for a total of four readings per condition (note: it has been recommended that auscultatory BP readings not be taken at more than 1.5-min intervals; see Papillo and Shapiro, 1990). Paced breathing was performed during each task and recovery period.

3.4. Quantification of dependent measures Standard cardiac time-domain indices, as well as frequency-domain measures derived from the autoregressive spectra, were calculated on the 4-min epochs. The measures calculated were (a) mean IBI (average of the IBIs measured from R-spike to R-spike in ms), (b) variance of IBI in ms2 (VAR), and (c) the average of the absolute values of successive differences in R–R intervals in ms (mean successive differences, M.S.D.), a measure that has been found to correlate highly with other indices of vagally mediated HR variability (Hayano et al., 1991). The autoregressive spectral analysis procedure used in our lab has been extensively described elsewhere and will not be described in detail here (Colombo et al.,

252

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

1990; Friedman et al., 1993b; Tyrrell et al., 1995; Thayer et al., 1996b). Briefly, the autoregressive program sequentially recognizes peaks of individual R-spikes from the ECG and calculates the autoregressive coefficients that define the power spectrum for that series. Power values are calculated in power spectral density units (ms2 Hz − 1) for each spectral component in the model that gives the best statistical fit. Components in the LF (0.039–0.15 Hz) and HF (0.18–0.35 Hz) ranges are then extracted. From these values, the LF/HF ratio is derived as an index of cardiac autonomic balance (Baselli et al., 1987; Malliani et al., 1990). Autoregressive spectral analysis has a number of advantages over classical Fourier analysis which include: (1) the ability to use the interval series directly; (2) the production of stable spectral estimates with relatively short data records; and (3) robustness to aperiodic influences such as arrhythmias (Pagani et al., 1986; Kay, 1988; Latson, 1994; Mainardi et al., 1995). From the full spectrum for each condition, the log of each frequency band was plotted against the log of the power obtained at each frequency. The slope of the linear regression of log-frequency on log-power was then calculated and served as a measure of spectral reserve. Finally, mean arterial blood pressure (MAP= diastolic pressure +1/3 pulse pressure; pulse pressure = systolic blood pressure− diastolic blood pressure), a measure of the average pressure during the cardiac cycle, was calculated from the average of four systolic and diastolic blood pressure readings per condition.

3.5. Data aggregation and analysis This experiment was concerned with individual differences (trait characteristics) in dynamic CV control mechanisms. To obtain accurate indices of CV traits, responses were aggregated across a variety of situations that induced varying levels of sympathetic and parasympathetic activation (see Manuck, 1994). Although this procedure provided valid estimates of CV traits, it is important to remember that, in the context of a stimulation study design, (1) these epochs were by intention quite different, and (2) the aggregation procedure, in combination with the pooled cross-sectional time series analysis, allowed for the estimation of the coupling parameters representative of the dynamic functioning of the system. As such, the tasks were not intended to reflect autonomic functioning in any one situation (West, 1990; Friedman et al., 1993a; Tyrrell et al., 1995). This approach is in line with recommendations that trait characteristics should be estimated through the aggregation of repeated measurements that are made over different situations (Hundleby et al., 1965; Nesselroade, 1990). Pooled time series designs were developed in the context of econometrics and have become popular in the analysis of biological data (Jaccard and Wan, 1993). These designs allow for the examination of both inter- and intra-subject variability in repeated measures experiments. In the present context, where the emphasis is on intra-individual variability within homogeneous groups, several approaches exist to minimize the within-group inter-individual variability, (Michela, 1990; Jaccard and Wan, 1993). The approach that we took was to standardize the data for each

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

253

individual before aggregating the data across individuals. In pooled-time series designs, some concern has been voiced with respect to statistical inference, especially in regard to the issue of the appropriate degrees of freedom for significance tests. A number of approaches have been suggested, many of which involve adjusting the degrees of freedom, not unlike the Greenhouse–Geisser adjustment for repeated measures analysis of variance (Vasey and Thayer, 1987). Thus, we used the very conservative lower bound suggested by Geisser and Greenhouse (1958). As such we report the uncorrected degrees of freedom with the understanding that division by the number of repeated measures per subject (seven in the present case) will yield the Geisser – Greenhouse lower bound estimate of epsilon. These corrected degrees of freedom are simply the number of independent units of observation (the number of subjects). Another point relevant to this concern is the exploratory versus confirmatory nature of the analyses (West and Hepworth, 1991). In this context, replication of a previous result is viewed as more important than the exactness of the standard errors associated with the test of significance. In the present experiment, all results remained statistically significant (most at their current level of significance), even after adjusting degrees of freedom down from the number of observations to the number of independent participants. Moreover, the estimation of the coupling parameters and the MANOVA results were explicit replications of previous findings. Thus, the results were reported with the degrees of freedom based upon the number of observations, but with full knowledge of the potential limitations and examination of appropriate alternatives.

4. Results Contrast and correlational analyses were performed with CSS software (Statsoft, 1987); nonparametric tests were conducted on the slope measure using Minitab (Minitab, 1987).

4.1. A priori contrasts In line with directional predictions based on data from Friedman et al. (1993b), as well as with the underlying conceptual model of the present study, one-tailed t-tests were performed on contrasts between blood phobics and panickers for all dependent measures except for slope. Two-tailed tests were used in contrasts involving the control group. Data were aggregated across experimental periods. Significant contrasts are shown in Table 1; no significant differences were found for LF or MAP. To summarize, the panickers had shorter IBIs, less MSD and HF spectral power, and higher LF/HF than blood phobics. The controls had longer IBIs, greater VAR and MSD, and more HF power than both panickers and blood phobics, and lower LF/HF ratios than the panickers. The trend across groups (from panic to control) is therefore towards increasing levels of vagally mediated HR variability.

254

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

4.2. Correlations Pearson correlations among all dependent measures were calculated to assess the coupling parameters between CV indices, and are displayed in Table 2. Correlation coefficients were calculated within subjects and across conditions, in accord with within-person correlational and pooled cross-sectional time series designs that have been advocated to capture the dynamic functioning of a system across diverse situations (Dielman, 1983; Michela, 1990). Data were standardized within subjects across tasks to avoid confounding between- and within-subject sources of variance. Several patterns of association observed by Friedman et al. (1993a) were replicated in the present study. A significant correlation between LF power and IBI was found in the panickers in the present study; this relationship was nonsignificant in the control group and in the blood phobic groups in both studies. The panickers also showed a significant relationship between LF and HF power in both studies, Table 1 Significant contrasts among panickers, blood phobics, and controls Variable

IBI (ms)

VAR (ms2)

MSD (ms)

HF power (ms2 Hz−1)

LF/HF

Panic (mean, S.D.) 761.8 (141.0)

3942 (4009)

44.4 (31.2)

991 (1225)

2.1(2.5)

P, panic; B, blood phobic; C, control.

Blood phobic (mean, S.D.)

Control (mean, S.D.)

837.1 (92.4)

905.2 (132.5)

4334 (2663)

55.6 (22.7)

6112 (4563)

71.4 (32.1)

1385 (1073)

2239 (1911)

1.3 (1.8)

1.0 (1.5)

T ratio, df, p value

PBB 4.59 (215) pB0.001 PBC 7.65 (214) pB0.001 BBC 4.30 (207) pB0.001 PBC 3.70 (214) pB0.001 BBC 3.44 (207) pB0.001 P =B N.S. PBB 3.05 (215) pB0.001 PBC 6.34 (214) pB0.001 BBC 4.11 (207) pB0.001 PBB 2.49 (212) pB0.01 PBC 5.67 (212) pB0.001 BBC 3.90 (203) pB0.001 PBB 2.41 (209) pB0.005 PBC 3.64 (203) pB0.001 B= C N.S.

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

255

Table 2 Correlations among CV variables for panickers, blood phobics, and controls IBI Panickers VAR 0.32*** MSD 0.52*** LF 0.25** HF 0.51*** LF/HF −0.25** MAP −0.03 Blood phobics VAR 0.13 MSD 0.60*** LF 0.09 HF 0.57*** LF/HF −0.26* MAP −0.21* Controls VAR 0.16 MSD 0.50*** LF 0.18 HF 0.36*** LF/HF −0.13 MAP −0.05

VAR

MSD

LF

HF

LF/HF

0.75*** 0.48*** 0.53*** −0.04 0.37***

0.32*** 0.76*** −0.29** 0.27**

0.20* 0.58*** 0.19*

−0.50c −0.26b

−0.07

0.56*** 0.59*** 0.41*** 0.26** 0.10

0.31** 0.89*** −0.20* −0.05

0.12 0.65*** −0.01

−0.42*** −0.01

0.03

0.63*** 0.48*** 0.34*** 0.15 0.16

0.35** 0.71*** −0.12 0.11

0.05 0.70*** 0.22*

−0.46*** 0.04

0.21*

*pB0.05; **pB0.01; ***pB0.001 (df = 203).

which was absent in the other experimental groups. Alternatively, the blood phobics were the only group in either study to display a significant association between MAP and IBI.

4.3. Spectral reser6e slopes As the distribution of spectral reserve slopes is not likely to be normal (Lipsitz et al., 1990), a nonparametric procedure (Wilcoxon Signed Rank Test) was used to compare groups on this measure. The panickers had significantly steeper slopes (median slope= −0.4736) than either the blood phobics (median slope= − 0.4265; W statistic =1698, n = 105, pB0.001) or controls (median slope= − 0.4035; W statistic =1432.5, n = 105, p B0.001). The difference in slope between blood phobics and controls was not significant. In order to use the slope in conjunction with overall variance, multivariate planned comparisons (from MANOVA) were conducted between the three groups, with slope and VAR as dependent measures. The use of these two indices in a multivariate manner allows for concomitant assessment of the totality of the HR variability (with VAR) and the sources of that variability (with slope). This approach represents a conceptual replication of the finding of Lipsitz et al. (1990), in which it was found that these two measures in combination were able to

256

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

distinguish the cardiovascular dynamics of their experimental groups. All three groups could be distinguished using this combination of variables. Blood phobics were significantly different from panickers (Wilk’s Lambda= 0.969 (2,204), pB 0.04) and controls (Wilk’s Lambda= 0.960 (2,189), pB 0.02). Controls and panickers were also significantly different (Wilk’s Lambda= 0.934 (2,190), p B0.002).

5. Discussion Overall, the substantive results from Friedman et al. (1993b) and Tyrrell et al. (1995) were replicated. Specifically, nonclinical panickers and blood phobics could be distinguished by their dominant mode of HR control as revealed by HR variability measures. Furthermore, these findings were extended to a comparison with a control group. With regard to vagal HR control, the panickers displayed the lowest level, blood phobics showed moderate amounts, and controls exhibited the most. Blood phobics were also characterized by an atypical linkage between cardiac rate and MAP. When used in conjunction with mean IBI variance, the slope measure of spectral reserve augmented these results by allowing for distinction among all groups. The control group showed more variance and spectral reserve than either anxiety group, but generally resembled the blood phobics more than the panickers in terms of HR variability and underlying autonomic control.

5.1. Anxiety and spectral reser6e The spectral reserve hypothesis amplifies the findings of Friedman et al. (1993a) and Tyrrell et al. (1995). The functional arrangement among the components derived from spectral analyses of the ECG and IBI time series reflects the responsivity of the CV system; loosely organized coupling allows the system to react with greater ease and flexibility. Adaptive biological systems are typically composed of loosely yoked bio-oscillators (West, 1990). Alternatively, both rigid coupling (strict periodicity) and complete decoupling (white noise) are the hallmarks of a disordered system (Goldberger and Rigney, 1990). Low spectral reserve has been implicated in a variety of cardiac arrhythmias (Goldberger and Rigney, 1990; Goldberger and West, 1987). Sympathetic dominance and reduced spectral resources may increase susceptibility to sinus tachycardia which has been found in ambulatory monitoring of panic attacks (Margraf, 1990; Taylor et al., 1986). The correlational data are consistent with the diminished spectral reserve that was found in the panickers. The significant association found between LF power and IBI in panickers is indicative of relatively substantial amounts of adrenergic HR control, which in turn is associated with reduced HR responsivity. Furthermore, the significant relationship between LF and HF power found in panickers suggests a more rigid coupling among spectral components and less spectral flexibility. The greater vagal control and spectral reserve found in the blood phobic group suggest a more adaptive CV system, but the significant negative association found

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

257

between IBI and MAP may reflect a tighter linkage between cardiac rate and hemodynamic regulation. This relationship implies that a decrease in BP may be accompanied by a decrease in HR, which is consistent with the mechanism of vasovagal syncope. In concert with vagal dominance, this arrangement may contribute towards vulnerability to syncope.

5.2. The state space of anxiety A nonlinear dynamics model that has been applied to psychopathology is heuristically useful in the present context (Globus and Arpaia, 1992). This representation depicts pathological conditions as ‘attractors’ in the ‘state space’ of an organism. Attractors are states that have a relatively high probability of occurrence in the overall psychobiological topography of an individual. Panic may be viewed as an attractor with a wide ‘basin of attraction’ in PD: i.e. a broad and ill-defined range of conditions that can generate a panic attack. This basin widens to encompass all stimuli present during an attack, consistent with portrayals of poor stimulus discrimination in conditioning models of panic (Wolpe and Rowan, 1988). The fear of recurrent panic attacks under a wide range of conditions may then lead to the development of agoraphobia, which is strongly associated with PD (DSMIV). Thus, panic can lead to an extreme limitation on behavioral flexibility. An analogy may be drawn between this restriction and the diminished physiological variability that has been reported in panic. A similar conceptual parallel has been made between physiological and cognitive rigidity in generalized anxiety disorder (Hazlett et al., 1994). Furthermore, such notions are consistent with the systems principle of similarities in process across multiple levels of analysis in nature (Mandell and Schlesinger, 1990), a phenomenon that has been specifically noted in anxiety (Kandel, 1983). Finally, low HR variability and cardiac vagal tone may directly reflect information processing components of anxiety by serving as indicators of central nervous system regulation of environmental engagement (Porges, 1991, 1992). Blood phobia presents far fewer constraints than panic. The basin of attraction for syncope is narrow and highly specific, so blood phobics are capable of a much greater range of behavioral flexibility than panickers. Indeed, blood phobics are not more prone to syncope in other (nonphobic) fear-inducing situations (Connolly et al., 1976). Consistent with the ‘self-similarity’ model, the blood phobics also displayed greater levels of vagally mediated HR variability than the panickers. Individuals who are prone to neither panic attacks nor blood phobic fainting episodes have even more available behavioral options. Furthermore, the control group in the present study displayed the greatest levels of HR variability and spectral reserve. However, the only defining psychological characteristics of this group were the absence of panic attacks and phobic reactions to blood. Future research should seek to explicate specific traits in nonclinical samples that may be related to HR variability.

258

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

5.3. Excess lability and autonomic flexibility When coupled with previous findings of reduced HR and EDA variability in anxiety, the present findings challenge models that posit excess lability and reactivity in anxiety. Such concepts persist in the current literature; e.g. excess physiological lability plays a role in both cognitive and conditioning models of panic (Wolpe and Rowan, 1988; Anastasiades et al., 1990). However, contemporary models of stability in biological systems question these traditional notions of ANS reactivity. Reactivity may be viewed as the integrity of organismic responsiveness to change, and includes qualities such as flexibility and resiliency. These traits have been found to be associated with increased levels of vagally mediated HR variability across a wide range of psychophysiological research domains. In this context, ‘reactivity’ has decidedly different connotations than the negative overtones it carries in either excess reactivity models of anxiety or CV disease (e.g. Manuck, 1994). In sum, the present study replicates and extends previous findings of reduced autonomic flexibility in chronic anxiety. The data also suggest that various forms of anxiety have distinct autonomic underpinnings. The analysis of HR variability appears to have particular theoretical, heuristic, and etiological utility in understanding the phenomenology of anxiety. This line of investigation resonates with contemporary biological systems models and provides an alternative to the orthodox homeostatic view of reactivity. Finally, the framework of autonomic flexibility suggests intriguing parallels across multiple levels of analysis in the biopsychology of anxiety.

6. Unlinked references AUTHOR, PLEASE CITE REFERENCE Hoehn-Saric and McLeod, 1988 IN TEXT

Acknowledgements Preparation of this paper was supported by National Institute of Health Training Grant HL-07560 (BHF), and the Rosenzweig Post-Doctoral Fellowship in Personality and Idiodynamics (BHF). The authors would like to thank Drs. Karen A. Matthews and Saul Rosenzweig for their support in the preparation of the manuscript, Dr. Thomas Borkovec for his guidance, and Mark Cheek, Eric Muth, and Patricia Palichat for their assistance in data collection and quantification.

References American Psychiatric Association, 1987. Diagnostic and Statistical Manual of Mental Disorders. 3rd ed., revised. American Psychiatric Association, Washington, DC.

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

259

American Psychiatric Association, 1994. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. American Psychiatric Association, Washington, DC. Anastasiades, P., Clark, D.M., Salkovskis, P., Middleton, H., Hackman, A., Johnston, D.W., 1990. Psychophysiological responses in panic and stress. J. Psychophysiol. 4, 331 – 338. Appel, M.V., Berger, R.D., Saul, J.P., Smith, J.P., Cohen, R.J., 1989. Beat to beat variability in cardiovascular variables: Noise or music? J. Am. Coll. Cardiol. 14, 1139 – 1148. Barlow, D.H., 1988. Anxiety and its Disorders. Guilford Press, New York. Baselli, G., Cerutti, S., Civardi, S., Lombardi, F., Malliani, A., Merri, M., Pagani, M., Rizzo, G., 1987. Heart rate variability signal processing: A quantitative approach as an aid to diagnosis in cardiovascular pathologies. Int. J. Bio-Med. Comput. 20, 51 – 70. Berne, R.M., Levy, M.N., 1992. Cardiovascular Physiology. 6th ed. Mosby-Year Book, St. Louis, MO. Berntson, G.G., Cacioppo, J.T., Quigley, K.S., 1991. Autonomic determinism: The modes of autonomic control, the doctrine of autonomic space, and the laws of autonomic constraint. Psychol. Rev. 98, 459–487. Borkovec, T.D., Rachman, S.J., 1979. The utility of analogue research. Behav. Res. Ther. 17, 253 – 261. Cannon, W.B., 1929. Bodily Changes in Pain, Hunger, Fear, and Rage. 2nd ed. D. Appleton, New York. Cannon, W.B., 1939. The Wisdom of the Body. Norton, New York. Caranasos, G.J., 1974. Neurocirculatory asthenia. In: Eliot, R. (Ed.), Stress and the Heart. Futura Publishing, Mount Kisco, NY, pp. 219 – 244. Carruthers, M., Taggart, P., 1973. Vagotonicity of violence: Biochemical and cardiac responses to violent films and television programmes. Br. Med. J. 3, 384 – 389. Colombo, R., Mazzuero, G., Soffiatino F., Ardizzoia M., Minuco, G., 1990. A comprehensive PC solution to heart rate variability analysis in mental stress. IEEE Comput. Cardiol. 475 – 478. Connolly, J., Hallam, R.S., Marks, I.M., 1976. Selective association of fainting with blood-injury-illness fear. Behav. Ther. 7, 8–13. Costello, C.G., 1971. Anxiety and the persisting novelty of input from the autonomic nervous system. Behav. Ther. 2, 8–13. Cruz, M., Larsen, R.J., 1995. Personality correlates of individual differences in electrodermal lability. Soc. Behav. Personal. 23, 93–104. Davis, R.C., 1958. The domain of homeostasis. Psychol. Rev. 65, 8 – 13. Deering, W., West, B.J., 1992. Fractal physiology. IEEE Eng. Biol. Med. 11, 40 – 46. Dielman, T.E., 1983. Pooled cross-sectional and time series data: A survey of current statistical methodology. Am. Statist. 37, 111–122. Epstein, S., O’Brien, E.J., 1985. The person-situation debate in historical and current perspectives. Psychol. Bull. 98, 1097–1126. Eysenck, H.J., 1970. The Structure of Human Personality. Meuthen, London. Fraser, F., Wilson, R.M., 1918. The sympathetic nervous system and the ‘irritable heart of soldiers’. Br. Med. J. 2, 27–29. Friedman, B.H., Thayer, J.F. Autonomic balance revisited: Panic anxiety and heart rate variability. J. Psychosom. Res. (in press). Friedman, B.H., Thayer, J.F., Borkovec, T.D., 1993a. Heart rate variability in generalized anxiety disorder (Abstract). Psychophysiology 30, S28. Friedman, B.H., Thayer, J.F., Borkovec, T.D., Tyrrell, R.A., Johnsen, B.H., Colombo, R., 1993b. Autonomic characteristics of nonclinical panic and blood phobia. Biol. Psychiatry 34, 298 – 310. Friedman, B.H., Thayer, J.F., Tyrrell, R.A., 1996. Spectral characteristics of heart period variability in shock avoidance and cold face stress in normal subjects. Clin. Autonom. Res. 6, 147 – 152. Geisser, S., Greenhouse, S.W., 1958. An extension of Box’s results on the use of the F distribution in multivariate analysis. Ann. Math. Statist. 29, 885 – 891. Glick, G., Yu, P.N., 1963. Hemodynamic changes during spontaneous vasovagal reactions. Am. J. Med. 34, 42–51. Globus, G.G., Arpaia, J.P., 1992. Psychiatry and the new dynamics. Biol. Psychiatry 35, 352 – 364. Goldberger, A.L., 1990. Fractal electrodynamics of the heartbeat. Ann. NY Acad. Sci. 591, 402 – 409. Goldberger, A.L., 1991. Is the normal heartbeat chaotic or homeostatic? News Physiol. Sci. 6, 87 – 91.

260

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

Goldberger, A.L., Rigney, D.R., 1990. Sudden death is not chaos. In: Krasner, S. (Ed.), The Ubiquity of Chaos. American Association for the Advancement of Science, Washington, DC. Goldberger, A.L., West, B.J., 1987. Fractals in physiology and medicine. Yale J. Biol. Med. 60, 421–435. Goldberger, A.L., Bhargava, V., West, B.J., Mandell, A.J., 1985. On a mechanism of cardiac electrical stability: The fractal hypothesis. Biophys. J. 48, 525 – 528. Goldstein, I.B., 1965. Physiological responses in anxious women patients. Arch. Gen. Psychiatry 10, 382–387. Grunhaus, L., Gloger, S., Weisstub, E., 1981. Panic attacks: A review of treatments and pathogenesis. J. Nerv. Ment. Dis. 169, 608–613. Hayano, J., Sakakibara, Y., Yamada, A., Yamada, M., Mukai, T., Fujinami, T., Yokoyama, K., Watanabe, Y., Takata, K., 1991. Accuracy of assessment of cardiac vagal tone by heart rate variability in normal subjects. Am. J. Cardiol. 67, 199 – 204. Hazlett, R.L., McLeod, D.R., Hoehn-Saric, R., 1994. Muscle tension in generalized anxiety disorder: Elevated muscle tonus or agitated movement? Psychophysiology 31, 189 – 195. Hoehn-Saric, R., McLeod, D.R., 1988. The peripheral sympathetic nervous system: Its role in pathological anxiety. Psychiatr. Clin. North Am. 11. Hoehn-Saric, R., McLeod, D.R., Zimmerli, W.D., 1989. Somatic manifestations in women with generalized anxiety disorder. Arch. Gen. Psychiatry 46, 1113 – 1119. Hoehn-Saric, R., McLeod, D.R., Zimmerli, W.D., 1991. Psychophysiological response patterns in panic disorder. Acta Psychiatr. Scand. 83, 4 – 11. Huang, M.H., Wolf, S.G., Armour, J.A., 1995. Shortening of the QT interval of the EKG is associated primarily with increased ventricular contractility rather than heart rate. Integr. Physiol. Behav. Sci. 30, 5–11. Hundleby, J.D., Pawlik, K., Cattell, R.B., 1965. Personality Factors in Objective Test Devices: A Critical Integration of a Quarter Century’s Research. Robert R. Knapp, San Diego, CA. Jaccard, J., Wan, C.K., 1993. Statistical analysis of temporal data with many observations: Issues for behavioral medicine. Ann. Behav. Med. 15, 41 – 50. Jennings, J.R., McKnight, J.D., 1994. Inferring vagal tone from heart rate variability. Psychosom. Med. 56, 194–196. Kagan, J., Reznick, J.S., Snidman, N., Johnson, M.O., Gibbons, J., Gersten, M., Biederman, J., Rosenbaum, J., 1990. Origins of panic disorder. In: Ballenger, J.C. (Ed.), Neurobiology of Panic Disorder. Wiley, New York, pp. 71–78. Kamarck, T.W., Jennings, J.R., 1991. Biobehavioral factors in sudden cardiac death. Psychol. Bull. 109, 17–28. Kandel, E.R., 1983. From metapsychology to molecular biology: Explorations into the nature of anxiety. Am. J. Psychiatry 140, 1277– 1293. Kaplan, D., Furman, M.I., Pincus, S.M., Ryan, S.M., Lipsitz, L.A., Goldberger, A.L., 1991. Aging and the complexity of cardiovascular dynamics. Biophys. J. 59, 945 – 949. Katkin, E.S., 1975. Electrodermal lability: A psychophysiological analysis of individual differences in response to stress. In: Sarason, I.G., Speilberger, C.D. (Eds.), Stress and Anxiety, vol. II. Hemisphere, Washington, DC, pp. 141–176. Kay, S.M., 1988. Modern Spectral Estimation: Theory and Application. Prentice-Hall, Engelwood Cliffs, NJ. Kelsey, R.M., 1991. Electrodermal lability and myocardial reactivity to stress. Psychophysiology 6, 619–631. Lacey, J.I., 1967. Somatic response patterning and stress: Some revisions of activation theory. In: Appley, M.H., Trumbull, R. (Eds.), Psychological Stress: Issues in Research. Appleton-CenturyCrofts, New York, pp. 14–42. Lader, M.H., 1980. The psychophysiology of anxiety. In: van Praag, H., Lader, M.H., Rafaelson, O., Sachar, E. (Eds.), Handbook of Biological Psychiatry. Marcel Dekker, New York, pp. 225 – 247. Latson, T.W., 1994. Principles and applications of heart rate variability analysis. In: Lynch, C. III (Ed.), Clinical Cardiac Electrophysiology: Perioperative Considerations. J.B. Lippincott, Philadelphia, pp. 307–349.

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

261

Levy, M.N., 1990. Autonomic interactions in cardiac control. Ann. NY Acad. Sci. 601, 209 – 221. Levy, M.N., Zieske, H., 1969. Autonomic control of cardiac pacemaker activity and atrioventricular transmission. J. Appl. Physiol. 27, 465 – 470. Lipsitz, L.A., Mietus, J., Moody, G.B., Goldberger, A.L., 1990. Spectral characteristics of heart rate variability before and during postural tilt. Circulation 81, 1803 – 1810. Lyonfields, J.D., Borkovec, T.D., Thayer, J.F., 1995. Vagal tone in generalized anxiety disorder and the effects of aversive imagery and worrisome thinking. Behav. Ther. 26, 457 – 466. Mainardi, L.T., Bianchi, A.M., Cerutti, S., 1995. Digital biomedical signal acquisition and processing. In: Bronzino, J.D. (Ed.), The Biomedical Engineering Handbook. CRC Press, Boca Raton, FL, pp. 828–852. Malik, M., Camm, A.J., 1990. Heart rate variability. Clin. Cardiol. 13, 570 – 576. Malliani, A., Lombardi, F., Pagani, M., Cerutti, S., 1990. Clinical exploration of the autonomic nervous system by means of electrocardiography. Ann. NY Acad. Sci. 601, 234 – 246. Mandell, A.J., Schlesinger, M.F., 1990. Lost choices: Parallelism and topological entropy decrements in neurobiological aging. In: Krasner, S. (Ed.), The Ubiquity of Chaos. American Association for the Advancement of Science, Washington, DC, pp. 35 – 46. Manuck, S., 1994. Cardiovascular reactivity in cardiovascular disease: ‘Once more unto the breach’. Int. J. Behav. Med. 1, 4–31. Margraf, J., 1990. Ambulatory monitoring of panic attacks. J. Psychophysiol. 4, 321 – 330. Mathews, A., 1990. Why worry? The cognitive function of anxiety. Behav. Res. Ther. 28, 455 – 468. Michela, J.L., 1990. Within-person correlational design and analysis. In: Hendrick, C., Clark, M.S. (Eds.), Research Methods in Personality and Social Psychology. Sage, Newbury Park, CA, pp. 279–311. Michelson, L., Mavissakalian, M., Marchione, K., 1988. Cognitive, behavioral, and psychophysiological treatments of agoraphobia: A comparative outcome investigation. AUTHOR, PLEASE GIVE FULL REFERENCE. Middleton, H.C., Ashby, M., 1995. Clinical recovery from panic disorder is associated with evidence of changes in cardiovascular regulation. Acta Psychiatr. Scand. 91, 108 – 113. Mitchell, J.H., Shapiro, A.P., 1991. The confusing syndromes of cardiovascular autonomic imbalance: Cardiovascular dysautonomias. In: Shapiro, A.P., Baum, A. (Eds.), Behavioral Aspects of Cardiovascular Disease. Lawrence Erlbaum, Hillsdale, NJ, pp. 75 – 102. Muth, E.R., Thayer, J.F., Stern, R.M., Friedman, B.H., Drake, C., 1996. The electrogastrogram and heart rate variability. Manuscript submitted for publication. Neary, R.S., Zuckerman, M., 1976. Sensation seeking, trait and state anxiety, and the electrodermal orienting response. Psychophysiology 13, 205 – 211. Nesselroade, J.R., 1990. Adult personality development: Issues in assessing constancy and change. In: Rubin, A.I., Zucker, R.A., Emmons, R.A., Frank, S. (Eds.), Studying Persons and Lives. Springer, New York, pp. 41–85. Norton, G.R., Dorward, J., Cox, B.J., 1986. Factors associated with panic attacks in nonclinical subjects. Behav. Ther. 17, 239–252. Ost, L.G., Sterner, U., Lindahl, I., 1984. Physiological responses in blood phobics. Behav. Res. Ther. 22, 109–117. Pagani, M., Lombardi, F., Guzzetti, S., Rimoldi, O., Furlan, R., Pizzineli, P., Sandrone, G., Malfato, G., Dell’Orto, S., Piccaluga, E., Turiel, M., Baselli, G., Cerutti, S., Malliani, A., 1986. Power spectral analysis of heart rate and arterial pressure variabilities as a marker of sympatho-vagal interaction in man and conscious dog. Circ. Res. 59, 178 – 193. Pagani, M., Rimoldi, O., Malliani, A., 1992. Low-frequency components of cardiovascular variabilities as markers of sympathetic modulation. Trends Pharmacol. Sci. 13, 50 – 54. Papillo, J., Shapiro, D., 1990. The cardiovascular system. In: Cacioppo, J.T., Tassinary, L.G. (Eds.), Principles of Psychophysiology: Physical, Social, and Inferential Elements. Cambridge University Press, Cambridge, pp. 456–512. Peng, C.K., Buldyrev, S.V., Hausdorff, J.M., Havlin, S., Mietus, J.E., Simons, M., Stanley, H.E., Goldberger, A.L., 1994. Non-equilibrium dynamics as an indispensable characteristic of a healthy biological system. Integr. Physiol. Behav. Sci. 29, 283 – 293.

262

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

Pincus, S.M., Gladstone, I.M., Ehrenkranz, R.A., 1991. A regularity statistic for medical data analysis. J. Clin. Monit. 7, 335–345. Porges, S. W. (1991) Vagal tone: An autonomic mediator of affect. In: Barber, J., Dodge, K.A. (Eds.), The Development of Emotion Regulation and Dysregulation. Cambridge University Press, Cambridge, pp. 111–128. Porges, S.W., 1992. Autonomic regulation and attention. In: Campbell, B.A., Hayne, H., Richardson, R. (Eds.), Attention and Information Processing in Infants and Adults. Erlbaum, Hillside, NJ, pp. 201–223. Porges, S.W., 1995. Orienting in a defensive world: Mammalian modifications of our evolutionary heritage. A polyvagal theory. Psychophysiology 32, 301 – 318. Reich, W., 1982. The Bioelectrical Investigation of Sexuality and Anxiety. Farrar, Straus & Giroux, New York (original work published in 1939). Rosenzweig, S., 1956. Projective methods and psychometric criteria. Aust. J. Psychol. 8, 152 – 155. Roth, W.T., Telch, M.J., Taylor, C.B., Sacitano, J.A., Gallen, C.G., Kopell, M.L., McClenahan, K.L., Agras, W.S., Pfefferbaum, A., 1986. Autonomic characteristics of agoraphobia with panic attacks. Biol. Psychiatry 21, 1133–1154. Saini, V., Verrier, R.L., 1989. The experimental study of behaviorally induced arrhythmias. In: Schneiderman, N., Weiss, S.M., Kaufmann, P.G. (Eds.), Handbook of Research Methods in Cardiovascular Behavioral Medicine. Plenum, New York, pp. 51 – 67. Sartory, G., Olajide, D., 1988. Vagal innervation techniques in the treatment of panic disorder. Behav. Res. Ther. 26, 431–434. Saul, J.P., 1990. Beat-to-beat variations of heart rate reflect modulation of cardiac autonomic outflow. News Physiol. Sci. 5, 32–37. Schell, A.M., Dawson, M.E., Filion, D.L., 1988. Psychophysiological correlates of electrodermal lability. Psychophysiology 25, 619–632. Sostek, A.J., 1978. Effects of electrodermal lability and payoff instructions on vigilance performance. Psychophysiology 15, 561–568. Stallone, D.D., Stunkard, A.J., 1991. The regulation of body weight: Evidence and clinical implications. Ann. Behav. Med. 13, 220–230. Stampler, F.M., 1982. Panic disorder: Description, conceptualization, and implications for treatment. Clin. Psychol. Rev. 2, 469–486. Stein, P.K., Bosner, M.S., Kleiger, R.E., Conger, B.M., 1994. Heart rate variability: A measure of cardiac autonomic tone. Am. Heart J. 127, 1376 – 1381. Surwillo, W.W., Quilter, R.E., 1965. The relation of frequency of spontaneous skin potential responses to vigilance and to age. Psychophysiology 1, 272 – 276. Taylor, C.B., Sheikh, J., Agras, W.S., Roth, W.T., Margraf, J., Ehlers, A., Maddock, R., Gossard, D., 1986. Ambulatory heart rate changes in patients with panic attacks. Am. J. Psychiatry 143, 478 – 482. Thyer, B.A., Curtis, G.C., 1985. On the diphasic nature of vasovagal fainting associated with blood-illness-injury phobia. Pavlovian J. Biol. Sci. 20, 84 – 87. Thayer, J.F., Friedman, B.H., 1993. Assessment of anxiety using heart rate nonlinear dynamics. In: Ditto, W. (Ed.) Chaos in Biology and Medicine. Proceedings of the International Society for Optical Engineering, 2036, pp. 42–48. Thayer, J.F., Friedman, B.H., Borkovec, T.D., 1996a. Autonomic characteristics of generalized anxiety disorder and worry. Biol. Psychiatry 39, 255 – 266. Thayer, J.F., Peasley, C., Muth, E.R., 1996b. Estimation of respiratory frequency from autoregressive spectral analysis of heart period. Biomed. Sci. Instrument. 32, 93 – 99. Tyrrell, R.A., Thayer, J.F., Friedman, B.H., Leibowitz, H.W., Francis, E.L., 1995. A behavioral link between the oculomotor and cardiovascular systems. Integr. Physiol. Behav. Sci. 30, 46 – 67. Uijtdehaage, S.H.J., Thayer, J.F., 1989. Sympathetic-parasympathetic interactions in heart rate responses to laboratory stressors (Abstract). Psychophysiology 26, S63. van Ravenswaaij-Arts, C.M.A., Kollee, L.A.A., Hopman, J.C.W., Stoelinga, G.B.A., van Geijn, H.P., 1993. Heart rate variability. Ann. Intern. Med. 118, 436 – 447. Vasey, M.W., Thayer, J.F., 1987. The continuing problem of false positives in repeated measures ANOVA in psychophysiology: A multivariate solution. Psychophysiology 24, 479.

B.H. Friedman, J.F. Thayer / Biological Psychology 47 (1998) 243–263

263

West, B.J., 1990. Fractal Physiology and Chaos in Medicine. World Scientific, Teaneck, NJ. West, S.G., Hepworth, J.T., 1991. Statistical issues in the study of temporal data: Daily experiences. J. Personal. 59, 609–662. Wilson, K.G., Sandler, L.S., Asmundson, G.J.G., Ediger, J.M., Larsen, D.K., Walker, J.R., 1992. Panic attacks in the nonclinical population: An empirical approach to case identification. J. Abnorm. Psychol. 101, 460–468. Wolf, S., 1995. Dogmas that have hindered understanding. Integr. Physiol. Behav. Sci. 30, 3 – 4. Wolpe, J., Rowan, V., 1988. Panic disorder: A product of classical conditioning. Behav. Res. Ther. 26, 441–450. Wood, P., 1941. DaCosta’s syndrome (or effort syndrome). Br. Med. J. 1, 767 – 845.

.