Reliability of the 35% carbon dioxide panic provocation challenge
Introduction
Several authors have formulated criteria for an ideal model of panic (Guttmacher et al., 1983, Gorman et al., 1987, Uhde and Tancer, 1990). The 35% CO2 challenge meets most of these criteria. Panic attacks provoked by the challenge are similar to naturally occurring panic attacks (Griez et al., 1987). Both peripheral and central manifestations of anxiety are present (Perna et al., 1994a, Verburg et al., 1995). The main difference between naturally occurring panic attacks and those induced by the challenge is that the effects of the challenge are short-lived and easily reversible (Griez and Van den Hout, 1983).
The challenge has the potential of a trait marker: healthy first-degree relatives of panic disorder patients are vulnerable to the challenge (Perna et al., 1995d). The challenge also has the potential of a state marker: drugs that are effective in panic disorder also block the response to the challenge (Pols et al., 1991, Pols et al., 1996a, Perna et al., 1994b). The effect of drugs on the challenge outcome seems to preceed the clinical effect, and is already apparent after one week (Bertani et al., 1997). A clinical measure that is not effective, taking alprazolam on an if-needed basis, does not decrease CO2 induced anxiety (Pols et al., 1996b). This last finding, however, is disputed (Sanderson et al., 1994).
The procedure is safe. CO2 is a non-toxic agent and, in the two main centers where the challenge is used (Milan and Maastricht), thousands of tests have been performed without major adverse effects.
The specificity of the challenge for panic disorder is thoroughly investigated. The test appeared to be almost, but not absolutely, specific for panic disorder. In certain groups of patients, mainly those with specific situational phobia (Verburg et al., 1994a), some respond similar to panic disorder patients. On the other hand, a recent analysis showed that extremely high anxiety responses to the challenge could be very specific for panic disorder (Verburg et al., submitted). Also, the challenge differentiates well between patients with panic disorder and patients with obsessive–compulsive disorder (Griez et al., 1990a, Perna et al., 1995b), generalized anxiety disorder (Verburg et al., 1995), major depression (Perna et al., 1995a) and animal phobia (Verburg et al., 1994a). The finding that vulnerability to this challenge is not absolutely specific for panic disorder is consistent with findings in lactate studies (Cowley and Arana, 1990).
The only criterion for an ideal model of panic that has not been investigated is reliability.
We already have some indications that the test outcome is reliable. The test results do not differ in different centers (Verburg et al., submitted). In a study that investigated the effects of Toloxatone on the response to 35% CO2, compared with placebo, Perna et al. (1994b)did not find attenuation of the CO2 response in the placebo group. One study may question the reliability of the challenge: the outcome of the test varies with the phase of the menstrual cycle (Perna et al., 1995c). This could be regarded as a drawback for the model. On the other hand, these findings could be regarded as an argument in favor of the validity of the challenge, because the clinical course of panic disorder is also influenced by hormonal changes. Pregnancy has a marked influence on the occurrence and frequency of panic attacks (George et al., 1987, Cowley and Roy-Byrne, 1989, Villeponteaux et al., 1992, Cohen et al., 1994, Verburg et al., 1994b). The menstrual cycle also appears to have an effect, although disputed, on the clinical course of panic disorder (Breier et al., 1986, Cameron et al., 1988, Stein et al., 1989, Cook et al., 1990, Kaspi et al., 1994).
To assess the reliability of the 35% CO2 challenge, we used a test–retest design, with an interval of 1 week between the two tests. We did not use an air-placebo condition in this study. We made this change in methodology for the following reasons: previous studies have sufficiently shown that placebo effects do not play an important role in 35% CO2 induced panic (Griez et al., 1987, Perna et al., 1994a). The use of an air-placebo condition has never led to any significant findings, neither did it ever influence the conclusion drawn from the results in the carbon dioxide condition. Also, this study presented us with methodological problems concerning the use of an air-placebo condition: subjects would have to be randomized into four groups (CO2 or air first in the first and second challenge), with a consequent loss of statistical power. And, more important, all subjects would know what to expect from the second inhalation at the second visit.
Therefore subjects performed one inhalation of a 35% CO2 mixture on both visits. The hypothesis was that the responses in anxiety on the Visual Analogue Scale for Anxiety (VAS-A) and the Panic Symptom List according to DSM-III-R (PSL-III-R) would be the same on both visits.
Section snippets
Subjects
Thirty panic disorder patients (11 men and 19 women) were included in this study. Twenty-four patients also met criteria for agoraphobia, seven suffered from a comorbid mood disorder (five from major depression and two from dysthymia). Patients were diagnosed according to DSM-III-R criteria (American Psychiatric Association, 1987). Diagnoses were made in consensus by two experienced clinicians. The diagnoses were confirmed using the Structured Clinical Interview for DSM-III-R (Spitzer et al.,
Drop-outs
Six patients dropped out after the first test. Five found the first test too frightening to do a second test. One patient had a panic attack in his car while driving home, which he attributed to the inhalation. These six patients were three men and three women, their mean age was 40.0 (S.D.=8.3). Five were agoraphobic, three suffered from a comorbid major depression. Their mean STAI score was 56.0 (S.D.=17.2). Before inhalation, they scored a mean of 38.3 (S.D.=39.3) on the VAS-A and 6.3
Discussion
The 35% CO2 challenge appears to have a good test–retest reliability. Post-CO2 scores on the VAS-A are reliable. PSL-III-R scores post-CO2 and delta-CO2 are also reliable.
The delta-CO2 scores on the VAS-A are unreliable. This seems to be a methodological problem rather than a problem related to the challenge itself. Notably, our results confirm the conclusions from an earlier study (Griez et al., 1990b) that baseline anxiety scores do not explain the level of anxiety induced by the challenge.
References (40)
- et al.
Menstrual fluctuation in the symptoms of panic anxiety
Journal of Affective Disorders
(1988) - et al.
Anxiety and the menstrual cycle in panic disorder
Journal of Affective Disorders
(1990) - et al.
Carbon dioxide and anxiety: Cardiovasculair effects of a single inhalation
Journal of Behavior Therapy and Experimental Psychiatry
(1983) - et al.
CO2 vulnerability in panic disorder
Psychiatry Research
(1987) - et al.
Specific sensitivity of patients with panic attacks to carbon dioxide inhalation
Psychiatry Research
(1990) - et al.
Pharmacological models of anxiety
Comprhensive Psychiatry
(1983) - et al.
Premenstrual exacerbation of symptoms in women with panic disorder
Journal of Anxiety Disorders
(1994) - et al.
35% CO2/65% O2 inhalation test in panic patients
Psychiatry Research
(1994) - et al.
35% CO2 challenge in panic and mood disorders
Journal of Affective Disorders
(1995) - et al.
Menstrual cyle-related sensitivity in 35% CO2 in panic patients
Biological Psychiatry
(1995)