Cognitive-behavior therapy for Japanese patients with panic disorder: Acute phase and one-year follow-up results
Each author's contribution: Y. Nakano is the primary investigator, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. K. Lee, Y. Noda, S; Ogawa, Y. Kinoshiota, T. Funayama, N. Watanabe, J. Chen, Y. Noguchi and T. A. Furukawa performed the clinical investigation (diagnosis, treatment, assessment and follow-up). TAF participated in the design of the study and supervised the overall conduct of the study. Y. Nakano wrote the first manuscript, all the authors took part in the rewriting of the manuscript, and all the authors approved the final manuscript.
*Yumi Nakano, MD, PhD, Department of Psychiatry and Cognitive-Behavioral Medicine, Nagoya City University Graduate School of Medical Sciences, Mizuho-cho, Mizuho-ku, Nagoya 467-8601 Japan. Email: firstname.lastname@example.org
Aim: The aim of this paper is to report the outcomes and follow-up data of our cognitive behavioral therapy program for Japanese patients with panic disorder and to examine the baseline predictors of their outcomes.
Methods: Seventy outpatients with panic disorder with or without agoraphobia were treated with manualized group cognitive behavioral therapy.
Results: Fourteen patients (20%) did not complete the program. Among the completers, the average Panic Disorder Severity Scale score fell from 12.8 at baseline to 7.1 post-therapy (44.7% reduction). This effectiveness was sustained for 1 year. While controlling for the baseline severity, the duration of illness and the baseline social dysfunction emerged as significant predictors of the outcome.
Conclusions: Our data suggest that group cognitive behavioral therapy for panic disorder can bring about as much symptom reduction among Japanese patients with panic disorder as among Western patients.
PANIC DISORDER (PD) with or without agoraphobia is a common psychiatric disease with a lifetime prevalence of approximately 3%.1,2 PD can lead to substantial reductions in social functions and quality of life.3,4 While the biological etiology of PD is similar to those of other anxiety disorders,5 it has a distinct psychological etiology; the process of catastrophic interpretation of anxiety-related physiological responses. Clinical intervention using cognitive behavioral therapy (CBT) for PD aims to alter this catastrophic cognition and to habituate patients to anxiety-provoking situations and anxiety-related sensations.6
Extensive evidence from randomized clinical trials (RCT) already exists to support CBT, medication, and combined CBT and medication therapy as efficacious treatments for PD with or without agoraphobia.7–9 These studies showed that the effectiveness of each treatment method was maintained for over 1 year. In Europe and North America, CBT and combined therapy have been reported to be feasible in daily clinical practices.10–12 However, whether comparable treatment outcomes can be achieved in non-Western clinical settings remains unproven. We translated into Japanese the group CBT program developed by the Clinical Research Unit for Anxiety and Depression at the University of New South Wales, Sydney, Australia,13 and have been conducting group CBT sessions based on this program for Japanese patients with PD since 2001.
The present study aims to report the outcomes and follow-up data of our CBT program and to examine the baseline predictors of the outcomes of Japanese patients with PD receiving CBT.
Seventy consecutive patients with PD were recruited into the CBT program at the Department of Psychiatry, Nagoya City University Hospital, Japan between June 2001 and July 2005. Some patients were referred from mental health professionals and others sought treatment for PD on their own and were self-referred. All the patients met the following entry criteria: (i) principal Axis 1 diagnosis of PD with or without agoraphobia according to the DSM-IV criteria, as assessed by the Structured Clinical Interview for DSM-IV;14 and (ii) absence of a history of psychosis or bipolar disorder, and of current substance-use disorder. Patients who had difficulty continuing their participation in the CBT program because of symptoms of other comorbid anxiety and/or depressive disorders were medicated with antidepressants and were admitted to the CBT program after their depression symptoms had abated and they still needed to and were able to attend the CBT sessions regularly.
The patients consisted of 24 men (34%) and 46 women, with a mean (±SD) age of 34.9 (±11.0) years. The mean baseline Panic Disorder Severity Scale (PDSS) score was 12.7 points, indicating a moderate level of PD symptomatology. The patients provided their written informed consent after full explanation of the objectives and procedures of the present study. The study's protocol was approved by the Ethics Committee of Nagoya City University Graduate School of Medical Sciences.
Our CBT program for the treatment of PD was based on the CBT program conducted at the Clinical Research Unit for Anxiety and Depression at the University of New South Wales, Sydney, Australia. We followed the published manual,13 which we had translated into Japanese. The CBT program comprised 10 sessions of lasting 2 hours each. The 1st to 9th sessions were held weekly, and the 10th session was conducted 4 weeks after the 9th session. The program included: (i) psychoeducation regarding the nature of anxiety and the causes of PD, (ii) breathing control training, (iii) cognitive restructuring, (iv) graded exposure to anxiety-provoking situations (as an assignment), and (v) graded exposure to panic-related physical sensations (in sessions and as an assignment). Components (i) and (ii) were conducted during the 1st session. Component (iii) was started in 3rd session, component (iv) in the 4th session, and component (v) in the 7th session. Patients were assigned some homework after every session to practice the components they had learned up to that session. The patients were treated in groups of three or four patients. Two therapists (one principal therapist and one co-therapist) led the group CBT sessions. The therapists who led the CBT sessions were psychiatrists or doctoral-level clinical psychologists with over 3 years' clinical experience. The senior psychiatrist (TAF), who had more than 15 years' experience as a clinician and had learnt the CBT program for PD at the Clinical Research Unit for Anxiety and Depression, trained the other therapists.
Patients were allowed to continue on antidepressants to control their anxiety and/or depression if they had been using antidepressants before beginning the CBT program and wished to continue doing so. Benzodiazepines are frequently used in Japan, even now, for the treatment of PD. However, other investigators have reported that benzodiazepines can reduce CBT effectiveness.15,16 Consequently, patients who had been using benzodiazepines regularly or occasionally during the daytime to control their anxiety were advised to taper and stop their usage or to switch to an antidepressant before starting the CBT. A few patients reported that they had not been able to stop taking anxiolytic benzodiazepine by the start of their CBT sessions. Medications for symptoms other than anxiety were not restricted, so that some patients used benzodiazepine hypnotics for the treatment of insomnia.
Before beginning the program, the clinicians in charge of the CBT group administered a semistructured diagnostic interview, Structured Clinical Interview for DSM-IV,14 in order to ascertain PD diagnosis and any comorbidities. In addition, one interview-based and seven self-reported questionnaires to measure various aspects of PD symptomatology, functional impairment and quality of life, broader psychopathology in general, and premorbid personality, were administered at the same time.
Panic Disorder Severity Scale
The PDSS17 is an interview-based, seven-item scale for assessing the overall severity of PD. The usual timeframe for the rating is the past month. The seven areas covered by the scale include: the frequency of the panic attacks, distress during the panic attacks, anticipatory anxiety, agoraphobic fear/avoidance, interoceptive fear/avoidance, impairment of work functioning, and impairment of social functioning. The reliability and validity of the Japanese version of the PDSS have been verified.18 Using the anchor-based approach, the following interpretative guidelines were utilized: among those with an established PD diagnosis, a total PDSS of 10 or less corresponded with a ‘mild’ PD, scores between 11 and 15 corresponded with a ‘moderate’ PD, and scores of 16 or greater corresponded with a ‘severe’ PD.18 This scale was used as the primary measure to assess the outcome of the CBT sessions.
The Fear Questionnaire – agoraphobia subscale
The Fear Questionnaire – agoraphobia subscale (FQ-A)19 is a self-reported instrument for measuring the severity of agoraphobic avoidance. Patients are asked to choose a number between 0 (would not avoid it) to 8 (always avoid it) to show how much he/she would avoid each of the listed situations because of fear or other unpleasant feelings. The items include five typical situations for agoraphobia. The possible score range is therefore 0–40. Good test–retest reliability and factor validity have been demonstrated.20
The Mobility Inventory for Agoraphobia
The Mobility Inventory for Agoraphobia21 is a self-reported instrument for measuring the severity of agoraphobic avoidance. Patients are asked to choose a number between 0 (never avoid) and 4 (always avoid) to show how they feel about 31 places and situations that they may avoid because of anxiety or phobia, either when they are alone (MI-AAL) or accompanied by someone else (MI-AAC). The average of all the items was taken as the scale score. The scale's reliability and concurrent and construct validity have been reported.21
Agoraphobic Cognitions Questionnaire and Body Sensations Questionnaire
The Agoraphobic Cognitions Questionnaire (ACQ)22 and the Body Sensations Questionnaire (BSQ)22 are self-reported questionnaires designed to measure aspects of ‘fear of fear’ observed in patients with PD: the former assesses maladaptive cognitions concerning potential harm of a panic attack, and the latter focuses on exaggerated fear responses touched off by sensations associated with anxiety. Each questionnaire contains 14 and 17 items, respectively, and is rated using five possible grades. The possible score ranges are therefore 0–56 in ACQ and 0–68 in BSQ. Both instruments have been shown to have good reliability and discriminant and construct validity.22
Symptom Checklist-90 Revised
The Symptom Checklist-90 Revised23 is a widely used assessment tool for general psychopathology. It contains 90 items, subdivided into nine subscales of somatization, obsessive-compulsive, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psychosis. The Japanese version has been tested in a general population.24 Each item is scored between 0 (not at all) and 5 (extremely), and the average of the relevant items was taken to be the subscale score.
Work, Home Management, Social and Private Leisure Activities Scale
The Work, Home Management, Social and Private Leisure Activities Scale (WHLS)19 is a self-reported instrument for measuring the severity of functional impairments at work, home, and during social and private activities. The patients' answers are graded between 0 (not at all impaired) to 8 (very severely impaired). Satisfactory reliability and construct validity have been reported.25
NEO Five Factor Index
The NEO Five Factor Index (NEO-FFI)26 is a 60-item self-reported questionnaire designed to measure five major personality dimensions: neuroticism, extraversion, conscientiousness, openness and agreeableness.26 The Japanese version has been tested in a general population and its reliability and validity have been reported.27,28 Each personality dimension was rated between 0 and 48.
The same instruments, except for the NEO-FFI, were repeated at the end of the program. The program completers were asked to fill out these rating scales, except for the PDSS and NEO-FFI, at 3 and 12 months after CBT and to send back the completed results by mail.
All the data were examined using SPSS 11.5J for Windows (SPSS Inc, Chicago, IL, USA). First, we used t-tests or χ2 tests to compare the demographic data, PD severity, current use of medication, and personality traits among the patients who completed the program and those who did not. Next, we estimated the outcome of the CBT program by calculating the percentage reduction in PDSS from baseline to post-therapy (= (baseline – post-therapy)/baseline PDSS × 100). The PDSS was our primary outcome. The statistical difference between the baseline and post-therapy PDSS scores was examined using a paired t-test. Third, to examine the baseline predictors of CBT outcome, we performed a multiple regression analysis using the baseline factors as independent variables and the post-therapy PDSS score as a dependent variable, while controlling for the baseline PDSS scores. Finally, to confirm that the outcome of the CBT was sustained, we used the Dunnett multiple comparison test to examine the follow-up data.
All the statistical tests were two-tailed, and an alpha value of less than 0.05 was considered statistically significant.
Completers and dropouts
Fourteen patients (20%) did not complete the CBT program. Table 1 shows the demographic characteristics and medication status of the patients at the start of the CBT program. The demographic and clinical characteristics did not differ between those who completed the program and those who did not.
Table 1. Demographics and baseline characteristics
|Sex (% male)||19 (33.9%)||5 (35.7%)||1.00|
|Mean age (SD)||35.9 (11.3)||30.9 (8.6)||0.078|
|Mean age at onset of panic disorder (SD)||28.9 (11.1)||25.9 (6.6)||0.20|
|Mean months of duration of panic disorder (SD)||81.4 (108.5)||55.6 (55.1)||0.23|
|Severity of panic disorder: baseline PDSS score (SD)||12.8 (4.7)||12.2 (4.8)||0.72|
|Agoraphobia||53 (94.6%)||13 (92.9%)||0.60|
| Depression||5 (8.9%)||2 (14.3%)||0.43|
| Other anxiety disorders||10 (17.9%)||4 (28.6%)||0.29|
| Specific phobia||2||0|| |
| Social phobia||2||2|| |
| OCD||2||0|| |
| PTSD||4||2|| |
| GAD||1||0|| |
|Social role functioning: baseline total WHLS score (SD)||10.7 (6.5)||10.5 (7.0)||0.92|
|Medication at start of CBT program|
| Antidepressants||34 (60.7)||8 (57.1)||1.00|
| Benzodiazepines||14 (25.0)||4 (28.6)||0.75|
| Neuroticism||25.2 (9.4)||28.4 (8.3)||0.24|
| Extraversion||26.0 (8.4)||27.4 (9.2)||0.61|
| Openness||27.4 (6.4)||31.0 (6.1)||0.083|
| Agreeableness||33.5 (7.5)||30.9 (4.9)||0.27|
| Conscientiousness||28.1 (8.8)||26.8 (8.2)||0.65|
Outcome of cognitive behavioral therapy program
The average PDSS score fell from 12.8 at baseline to 7.1 post-therapy (P < 0.001). This means that the mean severity of the PD decreased from a moderate level to a mild level. The mean percentage reduction in the PDSS total score was 44.7%. The effectiveness of the program seemed to increase with time and with experience on the part of treating clinicians, as the percentage reduction among the first 20 patients was 31.4% and that among the most recent 20 patients was 53.4%, while the dropout rate in the first and the most recent 20 patients was the same (15%). The scores for the rating scales related to PD symptoms are shown in Table 2.
Table 2. Rating scale scores for cognitive behavioral therapy completers (n = 56)
|PDSS||12.8 ± 4.70||7.05 ± 5.13||<0.001||<0.006*|
|FQ-A||13.9 ± 1.42||6.46 ± 1.05||<0.001||<0.006*|
|MI-AAC||0.94 ± 0.10||0.51 ± 0.08||<0.001||<0.006*|
|MI-AAL||1.66 ± 0.14||0.90 ± 0.11||<0.001||<0.006*|
|ACQ||14.5 ± 9.09||7.8 ± 6.62||<0.001||<0.006*|
|BSQ||29.7 ± 13.9||18.4 ± 12.7||<0.001||<0.006*|
Predictors of cognitive behavioral therapy outcome
First, a simple regression analysis of the baseline and post-therapy PDSS scores was conducted. The baseline PDSS score significantly predicted the post-therapy PDSS score. We therefore decided to control for the baseline PDSS score in subsequent analyses.
Second, a multiple regression analysis, using the baseline PDSS score and each baseline factor as independent variables and the post-therapy PDSS score as a dependent valuable, was conducted. This is a separate regression analysis for each predictor variable while controlling for baseline PDSS scores. Significant correlations between the post-therapy PDSS score and the following baseline factors were found: duration of illness, sequential group number, regular use of benzodiazepine, the FQ-agoraphobia score, the MI-accompanied score, the MI-alone score, and the WHLS total score (Table 3). Finally, to determine whether the identified predictors of the post-therapy PDSS score remained significant after controlling for the other variables, we performed a multiple regression analysis using five of the above mentioned factors (in order to avoid multicollinearity, the FQ-agoraphobia score and MI-accompanied score were excluded because they were strongly correlated with the MI-alone score) and the pretreatment PDSS total score. Only the duration of illness and the WHLS total score remained statistically significant. The multiple correlation coefficient was 0.73 in a stepwise multiple regression analysis.
Table 3. Relation between baseline variables and post-therapy Panic Disorder Severity Scale (results of regression analysis)
|Severity of panic disorder (PDSS)||0.39||0.003**|
|Age at onset of panic disorder||−0.27||0.034|
|Age at start of CBT program||0.13||0.31|
|Duration of illness before entering the CBT program||0.51||<0.001**|
| Other anxiety disorders||−0.11||0.40|
|Sequential group number||−0.29||0.022*|
|Severity of agoraphobic avoidance|
| MI accompanied||0.33||0.037*|
| MI alone||0.49||0.002**|
|Severity of automatic thoughts|
|Severity of fear of body sensations|
|Use of antidepressants||0.14||0.27|
|Use of benzodiazepine||0.23||0.074|
| Regular use||0.26||0.037*|
| Occasional use||0.22||0.082|
|Social Role Functioning (WHLS total score)||0.47||0.001**|
|Psychiatric symptoms (SCL-90-R)|
| Interpersonal sensitivity||−0.094||0.81|
| Phobic anxiety||0.51||0.11|
| Paranoid ideation||0.14||0.72|
Among the 57 completers, 39 patients (68%) completed all the self-reported questionnaires at pre- and post-treatment and at 3 and 12 months after CBT. Table 4 shows the changes in each score. The pretreatment scores were significantly higher than, but the post-treatment and 3-month follow-up scores were not significantly different from, the 1-year follow-up data for all the scales examined, as analyzed using the Dunnett test.
Table 4. Pre-treatment and 1-year follow-up rating scale scores
|FQ-A||13.9 ± 10.7||6.1 ± 7.8||5.4 ± 6.9||5.4 ± 7.1|
|MI AAC||0.9 ± 0.8||0.5 ± 0.6||0.5 ± 0.6||0.4 ± 0.6|
|MI AAL||1.7 ± 1.1||0.9 ± 0.9||0.8 ± 0.9||0.8 ± 0.9|
|ACQ†||13.3 ± 8.5||5.9 ± 5.2||5.8 ± 4.1||5.9 ± 5.5|
|BSQ||28.6 ± 13.6||15.1 ± 10.7||15.3 ± 11.0||15.2 ± 11.3|
|WHL†||11.8 ± 5.8||6.7 ± 6.3||6.9 ± 7.3||6.0 ± 6.4|
First of all, the present study demonstrated that a comprehensive CBT program for PD, originally developed in Western countries, is feasible and effective in a Japanese clinical setting.
The CBT program appears to be equally acceptable to Japanese patients with PD as to Western patients, because our dropout rates (20%) are comparable to those reported in Western studies. For example, the dropout rate in Wade et al.'s naturalistic study of CBT for PD from a community mental health center in USA was 26%,11 in Martinsen et al.'s report from the CBT program conducted in a regional hospital in Norway was 14%,10 and in the German study conducted in outpatient clinics was 22%.12 These dropout rates seem to be generally lower than those reported in RCTs. For example, in the active treatment arms of Barlow et al.'s landmark RCT of PD, the dropout rates were 38%.8
The reduction in PD symptomatology in our cohort also appears to be comparable to that observed in Western clinical settings. The CBT program at a community mental health center in the USA was able to reduce the FQ-agoraphobia subscale by 47%,11 while we obtained a 53% reduction. The group CBT program conducted in a regional hospital in Norway showed a mean of 41% reduction in the ACQ score,11 while our sample showed a mean of 47% reduction. Another naturalistic study from Germany also reported a similar reduction rate because their intensive exposure program showed a mean reduction of 46% in the ACQ score. The original program at the Clinical Research Unit for Anxiety and Depression in the University of New South Wales in Sydney, from which our program was developed, reported effect sizes 0.78 and 0.76 for ACQ and BSQ, respectively,29 while ours were 0.76 and 0.81.
Our treatment outcomes may be, however, slightly inferior to those reported in more controlled settings of RCTs. In the above mentioned RCT of CBT, imipramine and their combination for the treatment of PD, Barlow et al. defined CBT responders as those patients whose PDSS scores were reduced by more than 40%.8 Using this criterion, the proportion of responders among the randomized subjects was 49% in the CBT-alone group and 60% in the CBT-plus-antidepressants group. In the present study, the proportion of patients showing 40% or greater reduction in the intention-to-treat sample was 46% (32/70). Furthermore, these treatment gains appeared to persist at least up to 1 year post-treatment.
Second, the present study investigated the baseline predictors of the treatment outcomes and found the duration of illness before entering the CBT program and the social role functioning score to be significant, independent predictors of the CBT outcome, while controlling for the baseline severity. No previous study has shown that social dysfunction at baseline is a predictor of treatment outcome, independent of the baseline symptom severity. Our findings suggest that the severer the impairment to a patient's daily life, the less benefit he or she is likely to gain by receiving CBT. Further investigation of the relationships between each aspect of social dysfunction and the CBT outcome is needed.
In contrast, one previous study has shown duration of illness to be a predictor of outcomes of CBT for PD.30 In univariate analyses agoraphobic avoidance and regular benzodiazepine use were also significantly correlated with the PD outcomes but they were not retained in the final stepwise multiple regression analyses. Moderate correlations between the duration of illness and the severity of agoraphobia (r = 0.41, P = 0.002) and between the duration of illness and benzodiazepine use (r = 0.32, P = 0.01) may be able to explain why only duration of illness was retained. Several studies have reported that benzodiazepine use limits CBT effectiveness,15,16 but these studies might not have considered the duration of illness as a confounding variable. In fact, two RCTs in which combined benzodiazepine and CBT therapy was examined both showed that concomitant benzodiazepine use did not adversely affect the acute phase effectiveness of CBT, although it was debatable whether the effectiveness of CBT would be maintained after medication was stopped.7,31 In Michelson et al.'s study a high rate of relapse was observed after benzodiazepine medication alone.32
Possible shortcomings of the present study include the following. First, this was an open trial and we had no control group. We therefore had to rely on within-group comparison. Second, the clinicians in charge of the CBT program administered PDSS at pre- and post-treatment so that this rating was not blind or independent. However the patients' self-ratings showed effectiveness of similar magnitude across the broad range of PD rating scales. Third, the rules regarding medication in this study were not precise. We deem this unavailable in naturalistic clinical settings.
In conclusion, the present study confirmed the feasibility and effectiveness of CBT for PD, which was originally developed for English-speaking patients, in a Japanese clinical setting. It may be time for us to disseminate the program widely throughout Japan and make it accessible to more Japanese patients.
This study was supported by a Grant-in-Aid from the Ministry of Health, Welfare and Labour, Japan. We thank Ms Kozue Maki, Kaori Kobori and Kayoko Hirabayashi for clerical management of the study database.