Carbon dioxide inhalation is known to induce an emotion similar to spontaneous panic in Panic Disorder patients. The affective response to carbon dioxide in healthy subjects was not clearly characterized yet.
Sixty-four healthy subjects underwent a double inhalation of four mixtures containing respectively 0, 9, 17.5 and 35% CO2 in compressed air, following a double blind, cross-over, randomized design. Affective responses were assessed according to DSM IV criteria for panic, using an Electronic Visual Analogue Scale and the Panic Symptom List. It was demonstrated that carbon dioxide challenges induced a dose dependent negative affect (p<0.0001). This affect was semantically identical to the DSM IV definition of panic. Older individuals were subjectively less sensitive to Carbon Dioxide (p<0.05).
CO2 induced affectivity may lay on a continuum with pathological panic attacks. Consistent with earlier suggestions that panic is a false biological alarm, the affective response to CO2 may be part of a protective system triggered by suffocation and acute metabolic distress.
Citation: Griez EJ, Colasanti A, van Diest R, Salamon E, Schruers K (2007) Carbon Dioxide Inhalation Induces Dose-Dependent and Age-Related Negative Affectivity. PLoS ONE 2(10): e987. doi:10.1371/journal.pone.0000987
Academic Editor: Justin Harris, University of Sydney, Australia
Received: April 22, 2007; Accepted: August 29, 2007; Published: October 3, 2007
Copyright: © 2007 Griez et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The authors have no support or funding to report.
Competing interests: The authors have declared that no competing interests exist.
Intolerance to carbon dioxide in anxiety prone individuals has been widely documented –. When inhaling hypercapnic gasses, subjects diagnosed with Panic Attacks (PA) shortly sense an instant affect that closely replicates spontaneous panic . Hence Klein inferred that pathological PA's may be false biological alarms, resulting from neuronal misfiring in an evolutionarily evolved, CO2 driven oversensitive suffocation monitor .
In other words it was suggested that panic may be an inborn behavioural response to a metabolic distress. If so, panic must belong to the behavioural repertoire of healthy individuals, the hypersensitive alarm in PD subjects corresponding to a normoresponsive system in others. Accordingly, the very same mechanism firing false alarms in PD patients as a response to moderate CO2 intake, should be activated in healthy subjects following higher doses of CO2.
Here we demonstrate in healthy individuals that increasing concentrations of CO2 dose dependently induce a negative affect and that this affect is semantically identical to panic, as defined in current psychiatric nosology.
Materials and Methods
Sixty-four volunteers provided their informed consent to participate in the study. There were 33 males and 31 females, aged 35.8 (SD = 15.9) and 31.1 (SD = 14.4) years respectively.
All potential participants had a complete inventory of medical history and a physical examination. Inclusion criteria were 18 to 65 years of age and a good present and past physical and mental condition. The mental condition was assessed by a structured psychiatric interview (Mini International Neuropsychiatric Interview) performed by a psychologist who was not directly involved in the study. Exclusion criteria included a history of pulmonary or cardiovascular disease, the presence of hypertension (diastolic>100 mmHg; systolic>170 mmHg), cerebral aneurysm, pregnancy, epilepsy, excessive smoking (>15 cigarettes/day), use of adrenergic receptor blockers and use of psychotropic medication. A history of affective or anxiety disorders within a first-degree relative excluded participation. Participants were also excluded if they reported common specific fears or if there was any suspicion of history of Panic Attacks. The ethics committee of the Academic Hospital of Maastricht approved the study.
The inhalation apparatus and the general procedures used in our laboratory have been described elsewhere .
More specifically, the procedure consisted for each subject in a double inhalation of four mixtures containing respectively 0, 9, 17.5 and 35% CO2 in compressed air, following a double blind randomized design.
Subjects were instructed in the use of a mask with a demand valve for self-administration of medical gasses and told that they would take a double vital capacity breath of four different concentrations of CO2 in air, which, though being a harmless physiologic substance, may cause brief neurovegetative responses and arousal or anxiety, depending on the concentration. Subjects were asked to exhale to the maximum, to position the inhalation mask on their face and inhale their full capacity as quickly as possible. Next they were to empty their lungs and refill them immediately with gas, whereupon they had to hold their breath for 5 seconds before exhaling. All inhalations took place within one week, on four separate days but at the same time for each probant. Care was taken that each inhalation represented at least 80% of the subject's vital capacity.
Affective responses were assessed with strict reference to the DSM IV (APA, 2000 ), which refers to a PA as “a discrete period of intense fear or discomfort”, in which four (or more) out of a list of thirteen predefined symptoms develop abruptly and reach a peak within 10 minutes.
Accordingly (Fig. S1) we used an electronic visual analogue scale for affect (eVAAS). The eVAAS was programmed on a Compaq Tablet PC, TC1000, with a 21,0 cm×16,0 cm touch screen having a 1027×748 pixel resolution. The VAAS was a 20 cm×1 cm horizontal bar. Subjects had to mark their anxiety level by tipping on the bar with a stylus, which had a 1 mm diameter spherical tip. The top of the display was labelled “fear or discomfort”. The scale was anchored from 0, “no fear/discomfort at all”, to 100, “the worst imaginable fear/discomfort ”. This instrument has been validated for use during 35% CO2 challenges [Van Duinen, M.A., Rickelt, J., Griez, E.J.L.. Validation of the eVAAS. 2007 unpublished].
Panic Symptom List (PSL-IV) was used to evaluate panic symptomatology . It consisted in a questionnaire listing thirteen items, each item representing a DSM-IV panic symptom, to be rated on a five point scale, from 0 (absent) to 4 (very intense) (Fig. S1).
The eVAAS was presented at baseline immediately before inhalation, which was followed after CO2 by multiple instant assessments, in fact as many as possible, during 60 seconds. This allowed the computation of both a peak value and an area under the curve (AUC). The Panic Symptom List was administered one minute before and after each inhalation. The total PSL score was calculated for each assessment.
Statistical analysis was performed on the eVAAS peak values, obtained by subtracting the baseline from the maximum value, and PSL total scores, represented by delta scores (post-pre assessment).
A one-way Manova of repeated measures with eVAAS peak values as the dependent variable and dose (exposure to the four mixtures of CO2) as the within-subjects factor was used to investigate the affective response to the various CO2 mixtures. The same analysis was conducted with eVAAS AUC scores and PSL individual and total scores as the dependent variables.
AUC was calculated by the trapezoidal rule extrapolation method. The data were reanalysed using a repeated measures design with eVAAS peak values or PSL value as the dependent variable, age (below and above age 38) as a between-subjects factor and exposure to the four mixtures as the within-subjects factor. In both analyses, orthogonal polynomial trend contrasts were used to search for the presence of significant linear and/or quadratic trends in case of a significant “dose” effect.
Subjects were divided in “responders” and “non-responders”, according to conservative criteria, proposed by others in previous CO2 challenge studies . Following those criteria an arbitrary eVAAS peak score of 50 was used as threshold to identify the responders (mean of the eVAAS peak scored during AIR+2 SD). In addition responders should report at least one-point increase for at least four of the 13 PSL symptoms.
a) eVAAS: air vs 9% p = 0.44; air vs 17.5% and vs 35% p ≤ 0.001; 9% vs 17.5% and vs 35% p ≤ 0.0001; 17.5% vs 35% p≤0.0001. b) Younger versus older subjects: p<0.05
a) eVAAS AUC score: air vs 9% p = 0.91; air vs 17.5% and vs 35% p ≤ 0.005; 9% vs 17.5% and vs 35% p ≤ 0.001; 17.5% vs 35% p≤0.0001. b) Time course of Fear/Discomfort in a single subject after the double inhalation of 0%, 9%, 17.5%, 35% CO2 respectively.
a) PSL: air vs 9% p = 0.76; air vs 17.5% and vs 35% p ≤ 0.0001; 9% vs 17.5% and vs 35% p ≤ 0.0001; 17.5% vs 35% p ≤ 0.0001. b) Younger versus older subjects: p = 0.217
eVAAS peak values showed CO2-triggered affectivity to be dose dependent (p<0.001), displaying an increase with concentration, which fits both a significant linear (p<0.0001) and quadratic pattern (p<0.001) (Fig. 1 a). eVAAS AUC values were also dose related (p<0.0001) (Fig. 2 a,b) and exhibited a significant linear and quadratic pattern (p<0.0001). PSL data yielded similar results, with comparable dose-response relationship (p<0.0001) and mathematical pattern (p<0.0001) (Fig. 3 a).
Changes in individual PSL scores symptoms are presented in Fig. 4. As far as cognitive symptoms are concerned, a PSL score increase >1 in any of CO2 conditions was shown in 53% of the subjects.
The sum of the symptom scores, defining a cognitive dimension (“derealization-depersonalization” score+“fear of loosing control-going crazy” score) ,  is presented in Fig. 5. Looking to the cognitive symptoms separately (to which we added fear of dying), 48%, 26%, and 8% of the subjects reported an increase of PSL score respectively, in any of CO2 condition. A significant CO2 dose-dependent relationship was evidenced for derealization-depersonalization, and fear of loosing control-going crazy (p<0.0001), which displayed a linear (p <0.001) and quadratic pattern (p<0.05). A similar significant dose-effect was found for fear of dying (p<0.05), however it appeared to be overall a very rare symptom.
According to above defined criteria there were 4 (6%), 4 (6%), 9 (14%), and 24 (37%) “responders” in the 0, 9, 17.5 and 35% CO2 conditions respectively (p≤0.0001).
Dividing the subjects in an older (> 38 years) and younger group, analysis revealed a significant difference in eVAAS peak scores (p<0.05) (Fig. 1 b). eVAAS peak values in both age groups increase in a significant linear (p<0.001) and quadratic pattern (p<0.001) and run parallel. PSL scores displayed that age effect was not significant (p = 0.217) (Fig. 3 b).
Dividing the subjects by gender, no significant differences were found between males and females in any of the assessed parameters (eVAAS peak score, eVAAS AUC score, PSL–IV score).
The double breath challenge induced an instant affect with a negative valence. As rated on the eVAAS, the healthy volunteers experienced a significant sense of “fear” or “discomfort” while reporting substantial panic symptomatology on the PSL. Both eVAAS and PSL were strictly based on DSM-IV semantics. The picture as a whole was a mathematical function of CO2 intake.
It may therefore be inferred that a double breath of increasing concentrations of CO2 dose dependently induced a condition complying with the formal criteria of panic in current psychiatric nosology.
There has been a wealth of evidence showing that experimental hypercapnia triggers PA's in patients diagnosed with PD , , , and conditions which are closely related to PD , . In contrast, the same procedure failed to affect patients with other disorders , in particular those with Generalized Anxiety Disorder , , Obsessive-Compulsive Disorder , , Eating Disorders , Major Depression  and control groups of healthy volunteers. First-degree relatives of PD patients however share a significant degree of CO2 vulnerability , , . In fact, the liability to experience panic with CO2 exposure discriminates between individuals at high and low risk for PD .
Recent reports have suggested that healthy individuals breathing a low 7% concentration of CO2 may display signs of generalized anxiety , . Yet, the present results are the first to demonstrate that CO2 dose dependently activates a condition identical to panic in healthy volunteers, regardless of any constitutional predisposition to psychiatric pathology.
Did CO2 induce a true emotion? While formally meeting all the criteria of a PA according to modern psychiatric nosology, the CO2 induced state we observed in our healthy subjects may have been a “phenocopy” of panic, the amalgam of autonomic symptoms of hypercapnia and some resulting physical discomfort. Amongst the 13 PSL items, we therefore separately analyzed the specific cognitive symptoms of “derealization” and “fear of loosing control”. Several studies have identified these symptoms as belonging to a specific psychological/cognitive cluster on basis of factor analysis , . Our results show that both “derealization” and “fear of loosing control” were linked to the doses of inhaled carbon dioxide in a significant linear and quadratic pattern. Across the procedure, “derealization/depersonalisation” displayed more than one point increase (on a five point scale) in about half of the subjects, and “fear of loosing control” in about one fourth of them. Fear of dying did not belong to the cognitive dimension in Meuret and Cox's studies , , nevertheless, from a conceptual point of view, it may refer to an extreme type of emotion. In the present study “fear of dying” remained very infrequent. However, when reported, we noted a significant dose-response relationship. It should be born in mind that all subjects were in a safe laboratory environment, and all had received ample reassurance regarding the safety of the intervention trough the informed consent procedure. This obviously influenced the psychological impact of CO2. Yet, modest as they are, cognitive shifts did occur, they were a function of the experimental procedure, and their occurrence was statistically significant.
This lends support to the idea that, beyond a particular threshold, carbon dioxide may yield genuine psychotropic properties in healthy individuals.
Influential authors have increasingly referred to emotions as evolutionarily derived, “organism-ready solutions” to face major survival problems , as brain representations of internal body states , and more specifically, as images of the “material me” arising from “the homeostatic condition of each individual's body” . The idea that panic may proceed from a suffocation alarm disrupted by acute CO2 loading is perfectly consonant with such views. Several pieces of evidence point to a connection between hypercapnia and emotion. For instance, it appears that central chemosensitivity is not restricted to medullary respiratory neurons. Severson and co-workers  have shown that midbrain raphe serotonergic neurons are CO2 sensors, and midbrain neurons are not believed to have any direct function in the control of ventilation. Instead these midbrain chemosensors head mainly in the rostral direction. They have been proposed to participate in the homeostasis of the brain via non-respiratory responses to hypercapnia, including behavioural reactions as hyperarousal and anxiety . Liotti et al. have produced neuroimaging evidence linking directly CO2 inhalation with brain structures related to emotions . Following CO2 induced breathlessness, healthy volunteers displayed limbic and paralimbic activation, and neuronal firing in the affective brain correlated with the sense of suffocation. The authors comment that this neuronal activity may reflect a primal emotion, in other words “a compelling interoceptor-driven affect, rooted in metabolic distress, and aimed at signalling that the existence of the organism is endangered.” In an earlier study on the characteristics of CO2 induced responses, healthy volunteers spontaneously described their subjective experience as “frightening”, “panicky” or “scaring”, while authors noted that the sensitivity of the feeling, which was poorly correlated with the ventilatory response, varied threefold among individuals .
Our study shows that CO2 intake induces an affective state, which is similar to the psychiatric picture of panic. Within subjects, we observe a significant interaction between the intensity of the affective response and the CO2 concentration of the inhaled mixture.
We show older subjects to display less behavioural vulnerability to CO2, compared to younger individuals. To the extent that CO2 intake is a valid model of panic, this difference between younger and older subjects strikingly evokes the decline of natural PA's and the progressive blunting of panic symptomatology in PD patients when they grow older . Most studies have found a lower prevalence of PD amongst elderly people . The decreased CO2 susceptibility in the elderly revealed by the present data reminds of a similar age effect found with experimental cholecystokinin provocation of panic . If midbrain serotonergic chemosensors are at work in the chain of events leading from CO2 to panic, the phenomenon observed in our study might be related to an age dependent decline of serotonergic activity , .
No gender differences were found. This is somewhat surprising in view of all epidemiological data showing women to be at greater risk for panic than men . Yet, a recent study in a nonclinical population shows women reporting more fear and panic than men after CO2 administration . Interestingly, when asked to rate their experience on a “like or dislike” dimension (which dimension has a conceptual overlap with “discomfort”), the gender difference disappeared. This suggests women being more prone than men to report a feeling as “anxiety”. Therefore, lumping together anxiety and discomfort in our eVAAS may have blunted a gender effect. It is noteworthy that the few existing reports about sex differences in the so-called condition “non fearful PD”, which diagnosis relies on “discomfort” rather than on “anxiety”, suggest that both genders have similar prevalence , .
A final comment applies to the potential of further work with CO2 challenges in healthy individuals. The panic model of CO2, in particular the single breath 35% CO2 procedure, has proven to be both valid and reliable . It has undergone extended pharmacological validation, e.g. Bertani, Perna et al. . Assuming that higher doses of CO2 activate the same physiologic chain of events in panic free individuals, CO2 challenges in healthy volunteers may have a strong potential as a substitute to early clinical trials in testing novel pharmacological compounds.
In conclusion, it appears that healthy individuals display a distinct behavioural vulnerability to increasing levels of acute hypercapnia. This effect is dose-dependent and shares a striking similarity with the psychiatric condition of panic.
CO2 susceptibility, sensed as acute affective distress may represent an evolutionarily evolved protective mechanism in case of impending asphyxia.
Experimental assessments. DSM IV TR criteria for Panic Attack; eVAAS for Fear/Discomfort; Panic Symptom List (PSL-IV)
(0.14 MB TIF)
Conceived and designed the experiments: AC EG KS. Performed the experiments: AC ES. Analyzed the data: AC EG Rv. Contributed reagents/materials/analysis tools: AC EG ES KS. Wrote the paper: AC EG Rv KS.
- 1. Papp LA, Martinez JM, Klein DF, Coplan JD, Norman RG, et al. (1997) Respiratory psychophysiology of panic disorder: three respiratory challenges in 98 subjects. Am J Psychiatry 154: 1557–1565.
- 2. Griez E, Schruers K (1998) Experimental pathophysiology of panic. J Psychosom Res 45: 493–503.
- 3. Coryell W, Pine D, Fyer A, Klein D (2006) Anxiety responses to CO2 inhalation in subjects at high-risk for panic disorder. J Affect Disord 92: 63–70.
- 4. Coryell W (1997) Hypersensitivity to carbon dioxide as a disease-specific trait marker. Biol Psychiatry 41: 259–263.
- 5. Klein DF (1993) False suffocation alarms, spontaneous panics, and related conditions. An integrative hypothesis. Arch Gen Psychiatry 50: 306–317.
- 6. Verburg K, Perna G, Griez EJL (2001) A case study of the 35% CO2 challenge. In: Griez EJL, Faravelli C, Nutt D, Zohar J, editors. Anxiety Disorders. An Introduction to Clinical Management and Research, Wiley, Chichester. pp. 341–357.
- 7. American Psychiatric Association (2000) DSM-IV-TR Diagnostic and Statistic Manual of Mental Disorders, Fourth Edition, Text Revision. Washington, D.C.: American Psychiatric Association. ed. 4.
- 8. Schruers K, Klaassen T, Pols H, Overbeek T, Deutz NE, et al. (2000) Effects of tryptophan depletion on carbon dioxide provoked panic in panic disorder patients. Psychiatry Res 93: 179–187.
- 9. Poma SZ, Milleri S, Squassante L, Nucci G, Bani M, et al. (2005) Characterization of a 7% carbon dioxide (CO2) inhalation paradigm to evoke anxiety symptoms in healthy subjects. J Psychopharmacol 19: 494–503.
- 10. Meuret AE, White KS, Ritz T, Roth WT, Hofmann SG, et al. (2006) Panic attack symptom dimensions and their relationship to illness characteristics in panic disorder. J Psychiatr Res 40: 520–527.
- 11. Cox BJ, Swinson RP, Endler NS, Norton GR (1994) The symptom structure of Panic Attacks. Compr Psychiatry 35: 349–353.
- 12. Caldirola D, Perna G, Arancio C, Bertani A, Bellodi L (1997) The 35% CO2 challenge test in patients with social phobia. Psychiatry Res 71: 41–48.
- 13. Verburg C, Griez E, Meijer JA (1994) 35% carbon dioxide challenge in simple phobias. Acta Psychiatr Scand 90: 420–423.
- 14. Griez E, Zandbergen J, Pols H, de Loof C (1990) Response to 35% CO2 as a marker of panic in severe anxiety. Am J Psychiatry 147: 796–797.
- 15. Verburg K, Griez E, Meijer J, Pols H (1995) Discrimination between panic disorder and generalized anxiety disorder by 35% carbon dioxide challenge. Am J Psychiatry 152: 1081–1083.
- 16. Perna G, Bussi R, Allevi L, Bellodi L (1999) Sensitivity to 35% carbon dioxide in patients with generalized anxiety disorder. J Clin Psychiatry 60: 379–384.
- 17. Griez E, de Loof C, Pols H, Zandbergen J, Lousberg H (1990) Specific sensitivity of patients with panic attacks to carbon dioxide inhalation. Psychiatry Res 31: 193–199.
- 18. Perna G, Bertani A, Arancio C, Ronchi P, Bellodi L (1995) Laboratory response of patients with panic and obsessive-compulsive disorders to 35% CO2 challenges. Am J Psychiatry 152: 85–89.
- 19. Perna G, Casolari A, Bussi R, Cucchi M, Arancio C, et al. (2004) Comparison of 35% carbon dioxide reactivity between panic disorder and eating disorder. Psychiatry Res 125: 277–283.
- 20. Perna G, Barbini B, Cocchi S, Bertani A, Gasperini M (1995) 35% CO2 challenge in panic and mood disorders. J Affect Disord 33: 189–194.
- 21. Perna G, Cocchi S, Bertani A, Arancio C, Bellodi L (1995) Sensitivity to 35% CO2 in healthy first-degree relatives of patients with panic disorder. Am J Psychiatry 152: 623–625.
- 22. van Beek N, Griez E (2000) Reactivity to a 35% CO2 challenge in healthy first-degree relatives of patients with panic disorder. Biol Psychiatry 47: 830–835.
- 23. Bailey JE, Argyropoulos SV, Kendrick AH, Nutt DJ (2005) Behavioral and cardiovascular effects of 7.5% CO2 in human volunteers. Depress Anxiety 21: 18–25.
- 24. Panksepp J (1998) Affective neuroscience: The foundations of human and animal emotions. London: Oxford University Press.
- 25. Damasio A (1999) The feeling of what happens: Body and Emotion in the Making of Consciousness. New York: Harcourt Brace.
- 26. Craig AD (2003) Interoception: the sense of the physiological condition of the body. Curr Opin Neurobiol 13: 500–5.
- 27. Severson CA, Wang W, Pieribone VA, Dohle CI, Richerson GB (2003) Midbrain serotonergic neurons are central pH chemoreceptors. Nat Neurosci 6: 1139–1140.
- 28. Richerson GB (2004) Serotonergic neurons as carbon dioxide sensors that maintain pH homeostasis. Nat Rev Neurosci 5: 449–461.
- 29. Liotti M, Brannan S, Egan G, Shade R, Madden L, et al. (2001) Brain responses associated with consciousness of breathlessness (air hunger). Proc Natl Acad Sci USA 98: 2035–2040.
- 30. Banzett RB, Lansing RW, Evans KC, Shea SA (1996) Stimulus-response characteristics of CO2-induced air hunger in normal subjects. Respir Physiol 103: 19–31.
- 31. Sheikh JI, Swales PJ, Carlson EB, Lindley SE (2004) Aging and panic disorder: phenomenology, comorbidity, and risk factors. Am J Geriatr Psychiatry 12: 102–109.
- 32. Krystal JH, Leaf PJ, Bruce ML, Charney DS (1992) Effects of age and alcoholism on the prevalence of panic disorder. Acta Psychiatr Scand 85: 77–82.
- 33. Flint AJ, Koszycki D, Vaccarino FJ, Cadieux A, Boulenger JP, et al. (1998) Effect of aging on cholecystokinin-induced panic. Am J Psychiatry 155: 283–285.
- 34. Yamamoto M, Suhara T, Okubo Y, Ichimiya T, Sudo Y, et al. (2002) Age-related decline of serotonin transporters in living human brain of healthy males. Life Sci 71: 751–757.
- 35. Lawlor BA, Sunderland T, Hill JL, Mellow AM, Molchan SE, et al. (1989) Evidence for a decline with age in behavioral responsivity to the serotonin agonist, m-chlorophenylpiperazine, in healthy human subjects. Psychiatry Res 29: 1–10.
- 36. Gater R, Tansella M, Korten A, Tiemens BG, Mavreas VG, et al. (1998) Sex differences in the prevalence and detection of depressive and anxiety disorders in general health care settings: report from the World Health Organization Collaborative Study on Psychological Problems in General Health Care. Arch Gen Psychiatry 55: 405–413.
- 37. Kelly MM, Forsyth JP, Karekla M (2006) Sex differences in response to a panicogenic challenge procedure: an experimental evaluation of panic vulnerability in a non-clinical sample. Behav Res Ther 44: 1421–1430.
- 38. Fleet RP, Martel JP, Lavoie KL, Dupuis G, Beitman BD (2000) Non-fearful panic disorder: a variant of panic in medical patients? Psychosomatics 41: 311–320.
- 39. Bringager CB, Dammen T, Friis S (2004) Nonfearful panic disorder in chest pain patients. Psychosomatics 45: 69–79.
- 40. Verburg K, Pols H, de Leeuw M, Griez E (1998) Reliability of the 35% carbon dioxide panic provocation challenge. Psychiatry Res 78: 207–214.
- 41. Bertani A, Perna G, Arancio C, Caldirola D, Bellodi L (1997) Pharmacologic effect of imipramine, paroxetine, and sertraline on 35% carbon dioxide hypersensitivity in panic patients: a double-blind, random, placebo-controlled study. J Clin Psychopharmacol 17: 97–101.