232 Martinez et al.

© 2001 WILEY-LISS, INC.

DEPRESSION AND ANXIETY 14:232–237 (2001)

RESPIRATORY VARIABILITY IN PANIC DISORDER

J.M. Martinez, M.A.,* J.M. Kent, M.D., J.D. Coplan, M.D., S.T. Browne, B.A., L.A. Papp, M.D.,G.M. Sullivan, M.D., M. Kleber, Ph.D., F. Perepletchikova, B.A., A.J. Fyer, M.D., D.F. Klein, M.D.,

and J.M. Gorman, M.D.

Disordered breathing may play an important role in the pathophysiology ofpanic disorder. Several studies have now indicated that panic disorder patientshave greater respiratory variability than normal controls. In this study, weexamine baseline respiratory measures in four diagnostic groups to determinewhether greater respiratory variability is specific to panic disorder andwhether effective anti-panic treatment alters respiratory variability. Patientswith panic disorder, major depression, or premenstrual dysphoric disorder, andnormal control subjects underwent two respiratory exposures (5% and 7% CO2

inhalation), while in a canopy system. Panic disorder patients returned after12 weeks of either anti-panic medication or cognitive behavioral therapy, andwere retested. Normal control subjects were also retested after a period of 12weeks. Panic disorder patients had significantly greater respiratory variabilityat baseline than normal control subjects and patients with major depression.The premenstrual dysphoric patients also had greater variability than the nor-mal control group. Panic disorder patients who panicked to 7% CO2 inhala-tion had significantly greater baseline variability than panic disorder patientswho did not panic. Anti-panic treatment did not significantly alter baselinerespiratory variability. Our data suggest that increased respiratory variabilitymay be an important trait feature for some panic disorder patients and maymake them more vulnerable to CO2-induced panic. Depression and Anxiety14:232–237, 2001. © 2001 Wiley-Liss, Inc.

Key words: respiration; variability; panic disorder; fear; trait

Department of Psychiatry, Columbia University, and Depart-ments of Clinical Psychobiology and Therapeutics, New YorkState Psychiatric Institute, New York, New York.

Contract grant sponsor: MHCRC; Contract grant numbers: MH-30960, MH-41778; Contract grant sponsor: Senior Scientist Award;Contract grant number: NIH-00416; Contract grant sponsor: Inde-pendent Scientist Award; Contract grant number: MH-01397; Con-tract grant sponsor: Center for Neuroscience; Contract grantnumber: MH-58911-P50.

*Correspondence to: Jose Martinez, Department of Psychiatry,Columbia University, 1051 Riverside Drive, Unit 14, New York,NY 10032. E-mail: josemar@pi.cpmc.columbia.edu

Received for publication 6 June 2000; Revised 1 June 2001; Ac-cepted 4 October 2000

INTRODUCTIONPanic disorder (PD) has been linked to respiratory abnor-malities ranging from benign chronic hyperventilation withassociated hypocapnia to hypersensitive respiratory controlsystems [Sinha et al., 2000; Wilhelm et al., 2001a].

Our group has previously found, in two separate stud-ies, that panic disorder subjects have greater respiratoryvariability than comparison subjects. One study [Papp etal., 1997] found that female panic disorder subjects hadgreater within-subject standard deviations for tidal vol-ume and minute ventilation compared to female controlsat baseline in the laboratory. Another study [Martinez etal., 1995], which recorded ambulatory respiration ofpanic disorder subjects outside of the laboratory, foundgreater variability in minute ventilation during sleep andgreater variability of tidal volume during symptomaticperiods while awake. Other investigators have shown ir-regularity in breathing patterns in panic disorder subjectsat rest (particularly greater variability in tidal volume)[Wilhelm et al., 2000a; Abelson et al., 1996, 2001] andduring sleep [Stein et al., 1995]. Recent studies also dem-onstrated greater tidal volume instability in PD subjectsthan in patients with generalized anxiety disorder (GAD)

[Wilhelm et al., 2001a] and a greater frequency of sigh-ing in PD subjects [Wilhelm et al., 2001a,b; Abelson etal., 2001]. In total, these reports suggest an intrinsic ir-regularity in breathing, which may be due to instabilityin the autonomic control of respiration in panic disorder.

Research Article: Respiratory Variability in Panic Disorder 233

This report is part of a larger study investigatingthe respiratory physiology and behavioral responses ofpanic disorder patients to carbon dioxide inhalation[Kent et al., 2001; Gorman et al., 2001]. This reportfocuses on baseline respiration in an attempt to ad-dress three main questions: 1) can we replicate ourprevious baseline finding of increased respiratory vari-ability in panic disorder subjects?; 2) is respiratoryvariability specific to panic disorder subjects or does italso occur in other psychiatric disorders?; 3) doestreatment with medication or cognitive behavioraltherapy normalize or alter this irregular breathing pat-tern? To address these questions, we measured severalrespiratory variables (respiratory rate, tidal volume,and minute ventilation) with a respiratory canopy inpatients with panic disorder, major depression, pre-menstrual dysphoric disorder, and healthy controls.

METHODSTo address the issue of specificity, we included a

group of patients with major depression who generallydo not respond to agents that produce panic, such ascarbon dioxide, and a group of patients with premen-strual dysphoric disorder, who have been shown to besensitive to CO2 inhalation and lactate infusion, re-sponding with panic attacks [Kent et al., 2001; Harrisonet al., 1989; Sandberg et al., 1993]. Age, gender ratio,and n for each of the four diagnostic groups—panic dis-order (PD), major depression (MD), premenstrual dys-phoric disorder (PMDD), and normal controls (NC)—are reported in Table 1. The panic disorder patientswere age-matched to the normal controls (± 5 years)and for sex distribution.

Subjects were recruited by advertisement or referredby other mental health clinicians. Normal controls wererecruited by advertisement and by the Normal ControlUnit of the Mental Health Clinical Research Center atthe New York State Psychiatric Institute. A psychiatristor psychologist performed a diagnostic interview on allsubjects. All subjects were required to be physicallyhealthy and the three patient groups were defined byDSM-IV criteria. Patients underwent a second inter-view by a different clinician who administered theStructured Clinical Interview for DSM-IV (SCID).

Panic disorder patients were excluded if they metcriteria for current major depressive disorder, obses-

sive-compulsive disorder, or substance use disorder, orif they had a lifetime history of schizophrenia or bipo-lar disorder. Major depression patients could not havea history of an anxiety disorder, except for specificphobia, or have experienced a panic attack within 6months of testing. PMDD patients could not have ahistory of anxiety disorder, except for specific phobia,or have experienced a panic attack or met criteria formajor depression within 6 months of testing. The nor-mal controls underwent a structured interview usingthe Social Anxiety and Distress Scale (SADS) or SCIDin addition to the psychiatric interview. All subjects hadan electrocardiogram and physical examination, andblood and urine samples were taken for routine chemis-tries, thyroid function tests, complete blood count, uri-nalysis, urine toxicology screen, and pregnancy test.

Eligible subjects received a description of the proce-dure and signed informed consent. The InstitutionalReview Board approved the study. Subjects were re-quired to have been psychotropic drug-free for at least2 weeks before the day of testing. However, the use ofbenzodiazepines up to the equivalent of diazepam 10mg per day until 3 days before testing was permitted.Only one panic disorder subject annually took benzo-diazepines within 2 weeks of testing.

PROCEDUREThe experiment consisted of five 20-minute peri-

ods, in which respiratory parameters and behavioralassessments were recorded. The subject was asked tolie down, and his or her head was placed in a clearplastic canopy. The canopy was sealed, but the subjectcould see and hear and be seen and heard at all times.The subject was instructed on how to open the canopyby flipping a latch (no subject actually did this). Thesubject could also signal to have the procedure termi-nated at any time.

The 5 periods consisted of: 1) room air breathing, 2)5% CO2, 3) room air breathing, 4) 7% CO2, and 5)room air breathing. Subjects were not informed aboutthe timing or type of each period. A rater blinded to thesubject’s diagnosis administered the Acute Panic Inven-tory (API) [Dillon et al., 1987], Borg Scale of Exertion[Borg, 1982] (a measure of dyspnea), and a 10-pointanxiety scale. The scales were administered upon enter-ing the canopy and then after each of the five periods ofthe study. If the subject had a panic attack, or if the sub-ject indicated they wanted to stop the study, the scaleswere administered at that time. Panic attacks were as-sessed first by a rater blinded to the subject’s diagnosis,then by the subject, who pointed to a card to answerwhether a panic attack had occurred while the rater wasout of the room. This report will concentrate on thebaseline group respiratory differences during the firstroom air period.

Among the PMDD subjects, 4 were tested in theluteal phase of the menstrual cycle and 6 in the folli-cular phase.

TABLE 1. Age and sex distribution among the fourdiagnostic groups

Group N Age ± SD Males Females

PDa 49 32.16 ± 7.78 26 23NC 30 29.47 ± 7.38 18 12MD 21 34.33 ± 8.44 11 10PMDD 10 32.30 ± 6.55 10aPD, panic disorder; NC, normal controls; MD, major depression; PMDD,premenstrual dysphoric disorder.

234 Martinez et al.

RETESTAfter the first canopy test, patients were offered

treatment with either medication or cognitive behav-ioral therapy (CBT). CBT treatment consisted ofweekly individual visits with a psychologist who fol-lowed the method described by Barlow et al. [1988],called Panic Control Therapy. This treatment consistsof 11 sessions over 12 weeks and includes educationabout panic attacks, breathing retraining, cognitivetherapy, and interoceptive desensitization.

A psychiatrist prescribed medication treatment. Pa-tients were considered to be treatment responders ifthey had a Clinical Global Impressions (CGI) score of“much better” or “very much better.” Nine patientswere retested after 12 weeks of medication treatment,while still on medication. Three of the 9 were treatedwith imipramine (IMI). At the time these 3 patientswere retested, 2 were on an IMI dose of 200 mg/dayand 1 was on a dose of 250 mg/day. Six of the 9 patientswho were retested were treated with selective serotoninreuptake inhibitors (SSRIs). At the time when these 6patients were rated as responders, one was on sertraline200 mg/day, one was on venlafaxine 225 mg/day, onewas on fluoxetine 10 mg/day, one was on fluvoxamine150 mg/day, and two were on paroxetine, one on a doseof 40 mg/day, and the other on a dose of 10 mg/day.Following a treatment program designed to last 3months, panic disorder subjects were asked to repeatthe canopy study. Normal controls were asked to return3 months later to repeat the canopy study as well.

DATA ANALYSISRespiratory data, collected during 20 minutes of

room air breathing, consisted of continuous measures(every other breath) of respiratory rate (RR), tidal vol-ume (TV), and minute ventilation (MV). Within-sub-ject standard deviations (ws-sd) and Von Neumann[Von Neumann et al., 1941] values were calculated foreach subject for the 20 minutes of baseline breathing.Von Neumann values are calculated by taking themean square successive differences (MSSD) betweenvery other breath in the 20-minute baseline period.

Baseline respiratory variability. Univariate Analysisof Variance (UNIANOVA, SPSS V9.0) with 4 diagnosticgroups (PD, MD, PMDD, NC) was performed on thews-sd values for each of the 3 measures (RR, TV, MV).Within the context of univariate analysis of variance,post-hoc least square difference (LSD) tests were per-formed between the 4 diagnostic groups. Because of sig-nificant Levene tests for different variances betweengroups, nonparametric statistical testing (Kruskal-WallisANOVA) was performed on the MSSD values for allfour groups for the three respiratory variables. Signifi-cant results from the Kruskal-Wallis ANOVA were fol-lowed by post-hoc Mann-Whitney U tests. Given ourprevious findings of female panic disorder subjects hav-ing higher wd-sd for RR and MV than NC females[Papp et al., 1997], we repeated all tests for men andwomen separately.

Effect of treatment on respiratory variability. Arepeated measures analysis of variance (RM-ANOVA)was performed between test 1 (baseline) and test 2 (re-test) for each respiratory measure for PD treatmentresponders and NC subjects who repeated the canopystudy. To examine potential differences between themedication and CBT treatments, we performed a 2 ×2 RM-ANOVA on the PD treatment responders be-tween the two treatment groups (CBT, medication)for the two time points (baseline, retest).

Respiratory predictors of panic response. In or-der to identify predictors of panic response in the panicdisorder group, Mann-Whitney U tests were per-formed on MSSD values at baseline between subjectswho subsequently would go on to panic and those whodid not panic to 5% and 7% CO2 inhalation.

RESULTSThere were no significant age differences among

the four groups or differences in sex distributionamong the PD, NC, and MD groups (Table 1). Themean baseline RR, TV, and MV among the fourgroups were not significantly different (Table 2; thesedata also presented elsewhere [Gorman et al., 2001]).Of concern was whether patients who repeated therespiratory testing post-treatment differed from thosewho did not return for retesting; however, there wereno significant differences at the first test in any of therespiratory variables between those PD patients whodid and those who did not return for the second respi-ratory test.

Test 1 (baseline) results. Table 2 shows the over-all means, ws-sd, and Von Neumann values (MSSD)for the three respiratory measures (RR, TV, MV) foreach group at test 1 (baseline).

Significant differences in the univariate analysis ofvariance tests were found for the ws-sd for TV (F =3.54; P < 0.017; df = 3,106) and for MV (F = 4.15; P <0.008; df = 3,106). Post-hoc LSD comparisons showedthat PD subjects had significantly greater ws-sd forTV (P < 0.002) and for MV (P < 0.003) than NC sub-jects. PD subjects also had significantly greater ws-sdfor MV than MD subjects (P < 0.015).

Nonparametric statistical testing (Kruskal-WallisANOVA) for the MSSD values among the four groupsdemonstrated significant differences for TV (H = 12.86;df = 3; P < 0.005) and for MV (H = 11.86; df = 3; P <0.008). Post-hoc Mann-Whitney U tests (with Bonfer-roni correction), showed that the PD group had signifi-cantly greater MSSD values than the NC group for TV(Z = –3.495; P < 0.003) and for MV (Z = 3.323; P <0.005).

In testing for sex differences, Mann-Whitney Utesting (with Bonferroni correction) showed that thefemale PD group had significantly higher MSSD val-ues than female NC subjects for both TV and MV (Z= –3.79; P < 0.001 and Z = –3.06; P < 0.01 respec-tively). The PMDD subjects had significantly higher

Research A

rticle: Respiratory Variability in P

anic Disorder

235

TABLE 3. Test 1 (baseline) and test 2 (retest)*

Test 1 (baseline)

Respiratory rate Tidal volume Minute ventilation

Group Mean (bpm) ws-sd MSSD Mean (bpm) ws-sd MSSD Mean (bpm) ws-sd MSSD

PD n=22 15.48 ± 3.39 3.99 ± 2.22 36.81 ± 48.17 323.67 ± 163.43 162.41 ± 126.10 77,655 ± 103,691 4.49 ± 1.99 2.00 ± 1.29 10.62 ± 15.14NC n=15 16.56 ± 2.62 3.14 ± 1.37 35.38 ± 30.89 241.00 ± 119.56 80.21 ± 37.82 22,671 ± 27,690 3.79 ± 1.75 1.19 ± 0.50 4.21 ± 3.41

Test 2 (retest)

Respiratory rate Tidal volume Minute ventilation

Group Mean (bpm) ws-sd MSSD Mean (bpm) ws-sd MSSD Mean (bpm) ws-sd MSSD

PD n=22 15.92 ± 2.82 4.35 ± 1.65 28.19 ± 23.16 355.27 ± 141.56 181.22 ± 110.08 66,769 ± 79,613 5.29 ± 1.95 2.61 ± 1.31 13.48 ± 13.92NC n=15 16.76 ± 2.60 4.09 ± 2.07 27.15 ± 36.66 282.08 ± 111.09 115.34 ± 51.33 22,225 ± 17,499 4.70 ± 1.59 1.87 ± 0.66 4.77 ± 3.84

*For abbreviations, see Table 2.

TABLE 2. Test 1: Baseline values for mean, ws-sd, and MSSD of three respiratory measures among four diagnostic groups

Respiratory rate Tidal volume Minute ventilation

Group Mean (bpm) ws-sda MSSDa Mean (bpm) ws-sd MSSD Mean (bpm) ws-sd MSSD

PDb n=49 15.70 ± 3.34 4.59 ± 2.75 47.67 ± 60.10 360.84 ± 261.66 190.94 ± 184.73 94,825 ± 127,864 4.85 ± 2.65 2.36 ± 1.80 18.22 ± 25.60NC n=30 16.63 ± 2.68 3.58 ± 2.43 28.97 ± 26.33 245.5 ± 105.37 83.69 ± 42.51 20,718 ± 23,213 3.95 ± 1.84 1.37 ± 0.68 4.50 ± 3.41MD n=21 16.94 ± 4.14 4.35 ± 2.62 45.97 ± 55.4 301.0 ± 241.76 122.26 ± 135.76 78,429 ± 159,015 4.38 ± 2.14 1.46 ± 1.14 11.16 ± 21.2PMDD n=10 16.62 ± 2.12 4.03 ± 2.29 49.56 ± 55.9 246.3 ± 106.09 149.06 ± 153.55 83,996 ± 159,244 3.67 ± 1.53 1.52 ± 1.05 17.31 ± 28.5aws-sd, within-subject standard deviations; MSSD, mean square successive differences.bPD, panic disorder; NC, normal controls; MD, major depression; PMDD, premenstrual dysphoric disorder.

236 Martinez et al.

MSSD values for TV (Z = –3.10; P < 0.001) than thefemale NC subjects. No other sex comparisons weresignificant between groups; therefore, females ac-counted for most of the differences between groups.

Test 2 (retest) results. Of the 49 panic disordersubjects who entered the study, 24 did not return forretesting. Nine of these 24 never entered treatment,10 entered CBT treatment and dropped out before12 weeks, and 5 entered medication treatment anddropped out before 12 weeks. Only 4 of the 15 pa-tients who started treatment, but did not return forrepeat respiratory testing, were rated as respondersbefore they dropped out.

Of the 25 panic disorder subjects who repeated therespiratory testing after treatment, one had a panic at-tack during the room air baseline period and wasdropped from the analysis. Two other PD subjectswere considered treatment nonresponders and werenot included in the analysis. Of the 22 PD subjectswho responded to treatment and returned for a retest-ing, 13 patients chose CBT treatment and 9 chosemedication. Fifteen normal control subjects also un-derwent repeat respiratory testing (Table 3).

A RM-ANOVA comparing ws-sd values for test 1(baseline) and test 2 (retest) between the PD and NCgroups showed no significant difference in ws-sd forRR from test 1 to test 2. However, there was a signifi-cant Group difference for the ws-sd for TV (Group: F= 10.22; P < 0.003; df = 1,35), but no Time or Groupby Time interaction. For the ws-sd for MV we founda significant Group difference (F = 8.47; P < 0.006; df= 1,35), and a significant effect of Time (F = 7.07; P <0.012; df = 1,35), but No Group by Time interaction.Separate RM-ANOVAs performed on the MSSD val-ues for the 3 respiratory variables did not reveal sig-nificant differences between test 1 and test 2. Inaddition, separate RM-ANOVAs performed on theMSSD and ws-sd values for the 3 respiratory variablesfor the PD treatment responder group showed no sig-nificant differences between the medication and CBTtreatments.

Respiratory predictors of panic response. For7% CO2 inhalation, panic disorder subjects who wenton to panic had significantly greater baseline MSSDvalues for MV (Z = –2.00; P < 0.045; n = 44), andgreater ws-sd for MV (t = 2.23; P < 0.031; df = 43.0sve.) than PD non-panickers. For 5% CO2 inhalation,there were no significant differences between the PDpanickers and PD nonpanickers.

DISCUSSIONThe current findings are consistent with earlier re-

ports by our group and others [Wilhelm et al., 2001a,b;Abelson et al., 1996, 2001; Stein et al., 1995] that PDsubjects have greater baseline respiratory variabilitythan healthy comparison subjects. In our study samplethis was especially true among female PD subjects. In-vestigation of the specificity of this finding indicates

that increased respiratory variability may not be spe-cific to PD, in that the small number of PMDD sub-jects studied also had greater variability in theirbreathing than female control subjects. Although thisfinding should be interpreted cautiously due to thesmall sample size, it is consistent with other data dem-onstrating that PMDD patients have similar respira-tory and behavioral responses to PD patients whenexposed to either CO2 inhalation or lactate challenges[Harrison et al., 1989; Sandberg et al., 1993]. In total,these findings in PMDD subjects suggest that theneurobiology of PMDD may be more closely linkedto anxious, rather than depressive, disorders. In thisstudy, patients with MD were not distinguishablefrom healthy controls on any measure of respiratoryvariability.

Panic disorder subjects who went on to panic toCO2 demonstrated the greatest baseline respiratoryvariability. This respiratory variability may be a distin-guishing trait related to vulnerability to CO2-inducedpanic. Whether this reflects a primary abnormality ofrespiratory regulation, making this group more sus-ceptible to the trigger of heightened CO2 inhalation,or is simply a function of heightened baseline anxiety,remains to be clarified.

Neither treatment with CBT nor medication al-tered respiratory variability significantly. Thus, a re-duction in respiratory variability does not appear to benecessary for treatment response. This finding is con-sistent with the work of Abelson et al. [2001], whofound that irregular breathing patterns (particularlytidal volume) persisted despite cognitive or respiratorymanipulations. This was true despite a reduction inpanic rates secondary to the cognitive intervention.

The results of our study must be considered in thecontext of several methodological limitations. Our res-piratory analysis equipment was set up to record everyother breath, versus breath by breath respiration. Thisallowed us to process large volumes of data recordedover 20-min periods. Although this method results insome loss of information, the method was used consis-tently across all groups and allowed for the analysis of alonger time frame, and thus should not introduce bias.The relatively small number of returning subjects whorepeated the respiratory testing after 12 weeks limitsthe power of the study, and thus the interpretation ofthe retest data; however, our findings are consistentwith that of previous reports suggesting that increasedvariability in respiratory measures as quantified by thews-sd and Von Neumann (MSSD) statistics is a traitmarker of panic disorder.

Interestingly, we report a significant time effect forthe RM-ANOVA of the ws-sd for MV, as well as agroup effect, but no interaction. This reflects the factthat both PD and MD subjects demonstrated greatervariability in MV on repeat testing. One possible expla-nation for this finding is that all subjects experiencedmore anticipatory anxiety (potentially driving variabil-ity) during the repeat testing in anticipation of CO2 in-halation, perceived previously as anxiety-provoking.

Research Article: Respiratory Variability in Panic Disorder 237

In summary, our data suggest that increased respira-tory variability may be a trait marker of patients withpanic disorder, that patients with greater variability maybe most vulnerable to CO2-induced panic, and that res-piratory instability is largely resistant to treatment—whether it be medication or cognitive behavioraltherapy. Whether increased respiratory variability re-flects a primary dysregulation of central respiratoryregulatory mechanisms at the level of the midbrain orat upstream limbic structures remains to be clarified.A more comprehensive understanding of the role ofrespiratory regulation in the pathophysiology of panicdisorder will include further research in the areas ofcortical/limbic/brainstem interactions in the regula-tion of breathing, potential modulators of respiratorypatterns, and the function of unstable respiration inthe psychobiology of panic disorder.

REFERENCESAbelson JL, Randolph MN, Weg JG, Curtis GC. 1996. Repsiratory

psychophysiology and anxiety: cognitive intervention in thedoxapram model of panic. Psychosomatic Med 58:302–313.

Abelson JL, Weg JG, Nesse RM, Curtis GC. 2001. Persistent res-piratory irregularity in patients with panic disorder. Biol Psy-chiatry 49:588–595.

Barlow DH. 1988. Anxiety and its disorders: the nature and treat-ment of anxiety and panic. New York: Guilford Press.

Borg GAV. 1982. Psychophysical bases of perceived exertion. MedSci Sports Exerc 14:377–381.

Dillon D, Gorman JM, Liebowitz MR, Fyer AJ, Klein DF. 1987.The measurement of lactate-induced panic and anxiety. Psychia-try Res 20:97–105.

Gorman JM, Kent JM, Martinez JM, Browne ST, Coplan JD, PappLA. 2001. Physiological changes during carbon dioxide inhala-tion in patients with panic disorder, major depression, and pre-menstrual dysphoric disorder. Arch Gen Psychiatry 58:125–131.

Harrison WM, Sandberg D, Gorman JM, Fyer M, Nee J, Uy J,Endicott J. 1989. Provication of panic with carbon dioxide inha-lation in patients with premenstrual dysphoria. Psychiatry Res27:183–192.

Kent JM, Papp LA, Martinez RM, Browne ST, Coplan JD, KleinDF, Gorman JM. 2001. Specificity of panic response to CO2 inha-lation in panic disorder: a comparison with major depression andpremenstrual dysphoric disorder. Am J Psychiatry 158:58–67.

Martinez JM, Papp LA, Coplan JD, Anderson DE, Mueller CM,Klein DF, Gorman JM. 1995. Ambulatory monitoring of respi-ration in anxiety. Anxiety 2:296–303.

Papp LA, Martinez JM, Klein DF, Coplan JD, Norman RG, ColeR, DeJesus MJ, Ross D, Goetz R, Gorman JM. 1997. Respira-tory psychophysiology of panic disorder: three respiratory chal-lenges in 98 subjects. Am J Psychiatry 154:1557–1565.

Sandberg D, Endicott J, Harrison W, Nee J, Gorman J. 1993. So-dium lactate infusion in late luteal phase dysphoric disorder. Psy-chiatry Res 46:79–88.

Sinha S, Papp LA, Gorman JM. 2000. How study of respiratoryphysiology aided our understanding of abnormal brain functionin panic disorder. J Affect Disorder 61(3):191–200.

Stein MB, Miller TW, Larsen DK, Kryger MH. 1995. Irregularbreathing during sleep in patients with panic disorder. Am J Psy-chiatry 152:1168–1173.

Von Neumann J, Kent RH, Bellison HR, Hart BI. 1941. The meansquare successive difference. Ann Math Statistics 122:53.

Wilhelm FH, Trabert W, Roth WT. 2001a. Physiologic instabilityin panic disorder. Biol Psychiatry 49:596–605.

Wilhelm FH, Trabert W, Roth WT. 2001b. Characteristics ofsighing in panic disorder. Biol Psychiatry 49:606–614.