- Email: [email protected]
Phobic symptoms as predictors of nonresponse to drug therapy in panic disorder patients (a preliminary report)
JOURNAL
OF
AFFECTIVE DISORDERS
EUEVIER
Journal
of Affective
Disorders
33 (1995) 31-38
Phobic symptoms as predictors of nonresponse to drug therapy in panic disorder patients (a preliminary report) Bernhard R. Slaap *, Irene M. van Vliet, Herman G.M. Westenberg, Johan A. den Boer Department of Psychiatry,
Rudolf Magnus Institute for Neuroscience, Academic Hospital Utrecht, PO Box 85500, 3508 GA Utrecht, The Netherlands Received
4 May 1994; Accepted
5 September
1994
Abstract that predict nonresponse to drug therapy (brofaromine or fluvoxamine) were investigated in a sample of patients. We used a strict definition of nonresponse to find patients who did not respond at all after 12 weeks of treatment. Using this definition, 15 patients (32.6%) were considered nonresponders. Nonresponders had a higher score on the Blood-Injury subscore of the Fear Questionnaire (FQ) and more often had high scores on several FQ subscores, indicative of comorbid phobic symptoms. These variables were subsequently used to Factors
44 panic disorder
predict nonresponse. Keywords:
Panic disorder;
Prediction
of response;
Brofaromine;
1. Introduction
The principal management strategy for patients with panic disorder (PD) with or without agoraphobia (DSM-III-R; American Psychiatric Association, 1987) is drug therapy in combination with behaviour therapy. Various drugs have been shown to be effective in the treatment of PD: tricyclic antidepressants (for review, see Modigh, 1987, and Liebowitz, 1989), selective serotonin reuptake inhibitors (SSRIs; for review, see den Boer and Westenberg, 1990, and Westenberg and den Boer, 1994) and monoamine oxidase in-
* Corresponding
author.
Fax: (31) (30) 541844.
01650327/95/$09.50 0 1995 Elsevier SSDI 0165-0327(94)00070-O
Science
Fluvoxamine
hibitors (MAOIS; for review, see Tyrer and Shawcross, 1988). In all these studies, beneficial effects on the number of panic attacks were reported. However, not all patients respond to treatment: 20-40% of the patients are nonresponders. At present, it is unclear whether there are differences in psychological or biological characteristics of responders vs. nonresponders. Predicting response or nonresponse to antidepressant drugs would be very helpful since it takes - 4 weeks for these drugs to become clinically effective. There is scarce literature on predictors of response in PD. Compared with the vast amount of literature on the prediction of response in depression (see for, e.g., Joyce and Paykel, 19891, this area still seems relatively unex-
B.V. All rights reserved
32
B.R. Sap
et al. /Journal
of Affectit,e Disorder.\ 33 (1995) 31-38
plored in PD. Some predictors of response to drug therapy have been reported in the literature: (1) early improvement on the number of spontaneous panic attacks during treatment with alprazolam (Albus et al., 1990); (2) lower baseline scores on the Hamilton Anxiety Scale (HAS; Hamilton, 1959) (Liebowitz et al., 1986; Rosenberg et al., 1991; Scheibe et al., 1992; Woodman et al., 1994); (3) lower baseline scores on the Hamilton Depression Scale (HDS; Hamilton, 1967) (Rosenberg et al., 1991); (4) age > 40 (Woodman et al., 1994); and (5) lower baseline levels of phobic symptoms (Woodman et al., 1994). In the present study, we report on possible predictors for nonresponse to drug treatment in PD patients. The idea to study the nonresponders is not without precedent: there have been others who have tried to define clinical characteristics of nonresponders to, e.g., behaviour therapy (Emmelkamp and Foa, 1983; Emmelkamp and van den Hout, 1983). The aim of the present study is to look for clinical variables, measured at baseline, which may distinguish responders from nonresponders in a patient sample suffering from PD. The variables that differ significantly between the groups will be used to predict nonresponsc to drug therapy.
2. Methods 2 1. Design
For the present study, the data of two previously published studies were pooled. In these studies brofaromine, a selective and reversible MAOI, was compared with either fluvoxamine, an SSRI (van Vliet et al., 1994), or placebo (van Vliet et al., 1993). Both studies had a double-blind design and lasted 12 weeks. The psychometric scales were similar in both studies. By pooling the data, we obtained a data base of 60 patients. 29 patients were treated with brofaromine, 15 patients were treated with fluvoxamine and 16 patients received placebo. The 44 patients who received drug treatment have been studied in this paper. The patients treated with placebo were excluded from the study because we were inter-
ested predictors of nonresponse to active drug treatment. Possible predictors of nonresponse to placebo will be the subject of another article. The patients who received drug treatment were analysed as one group in this study. ANOVA with drug (brofaromine or fluvoxamine) as covariate showed no significant differences between the drug groups. Furthermore, from the study of van Vliet et al. (1994) it appears that there are no significant differences between brofaromine and fluvoxamine on any of the outcome measures. Nonresponders were defined as patients who still had panic attacks at endpoint and who had a reduction of < 50% on the Agoraphobia subscore of the Fear Questionnaire (FQ; Marks and Mathews. 1979). Patients who were free of panic attacks at endpoint and/or showed an improvement of 2 50% on the Agoraphobia subscore were considered to be responders. 2.2. Patients For each study, 30 patients were recruited from the Outpatient Clinic, Department of Biological Psychiatry, University Hospital, Utrecht, The Netherlands. The studies were approved by the Ethics Committee of the Academic Hospital and informed consent was obtained from all patients. Included in the studies were patients suffering from PD with Agoraphobia (59 patients) or without Agoraphobia (one patient) according to DSM-III-R criteria. Excluded were patients with other anxiety disorders, major affective disorder or psychotic disorder, alcohol abuse, severe sleep disturbance and patients suffering from other medical problems on the basis of a complete medical evaluation. Patients with a score of r 15 on the HDS were also excluded. During treatment, the use of other psychotropic drugs was not allowed except the use of oxazepam to a maximum of 20 mg daily if required. 60 patients, 54 females and six males were enrolled and treated for 12 weeks using a double-blind placebo controlled design. Mean age (*SD) was 36.6 f 7.1 years (range 20-50) and the mean duration of illness was 9.7 k 6.2 years (range 0.6-32). In each study, patients were randomly allocated to one of the two treatment groups: bro-
B.R. Slaap et al. /Journal
of Affective Disorders 33 (1995) 31-38
faromine or placebo in one study (van Vliet et al., 1993), or brofaromine or fluvoxamine in the other (van Vliet et al., 1994). 2.3. Symptom assessment At baseline and at the end of treatment, the HDS was completed. Efficacy of treatment was further assessed using the FQ and the HAS. The FQ and the HAS were completed on baseline and at the end of weeks 1, 2, 4 and 8 and at the end of week 12. The Utrecht Panic Inventory (UPI) was completed daily. This is a questionnaire, based on the DSM-III-R definition of a panic attack, measuring the frequency and intensity of panic attacks and the associated symptoms. Blood plasma levels of brofaromine or fluvoxamine were monitored to ensure compliance to treatment.
33
were computed which divided patients in high or low scorers on a FQ subscore. Patients were considered high scorers on a FQ subscore when they had a score of 2 20. (The maximum score on subscore is 40.) From normative data of the FQ in the article of Oei et al. (19911, it appears that approximately half of the patients with PD with agoraphobia had a score of 2 20 on the FQ Agoraphobia subscore (FQ-AG). N 30% of the patients had a score of 2 20 on the FQ Social Phobia subscore (FQ-SP) and N 25% had a score of 2 20 on the FQ Blood-Injury phobia subscore (FQ-BI). For the prediction of nonresponse to drug therapy, a logistic regression analysis was carried out for all significant baseline variables. Subsequently, a best-fitting model was made with all significant variables using logistic regression with backwards elimination. As criterion for variable selection, the likelihood-ratio test was used.
2.4. Data analysis Data were analysed using a commercially available statistical package (SPSS). To analyse the differences between responders and nonresponders, an ANOVA was performed on the baseline variables. When the assumption of nora nonparametric test mality was violated, (Kruskal-Wallis ANOVA) was applied. The categorical data were tested with the Fisher’s exact test (two-tail). To further investigate the influence of phobic symptoms on nonresponse, indicator variables
Table 1 ANOVA/Kruskal-Wallis
of psychometric
3. Results 3.1. Clinical variables
Of the 44 patients receiving medication (brofaromine or fluvoxaminel, there was one patient who did not complete the UPI, therefore, this patient was not entered into the statistical analysis. Of the remaining 43 patients, 15 patients (32.6%) were nonresponders, 11 patients (37.9%) of the brofaromine group and four patients
scales
Variable
Responder
HAS HDS FQ Total Score FQ-AG subscore FQ-SP subscore FQ-BI subscore
24.93 9.68 56.75 30.36 14.79 11.61
? + k f f +
(mean + SEMI
2.57 2.07 3.09 8.13 9.26 1.44
Nonresponder 25.27 10.80 66.07 30.67 17.53 17.87
+ + + + + k
4.27 2.11 4.25 7.35 8.54 1.37
UP1 week 1: number
of panic attacks
4.86 + 0.77
8.07 ? 1.39
UPI week 1: tingling
sensations
0.64 f 0.17
1.94 f 0.49
(mean f SEM)
P NS NS NS NS NS F = 8.00;df = 1,41; P = 0.0072 F = 4,87; df = 1.41: P = 0.0329 Kruskal-Wallis; P = 0.0199
34
B. R. Slaap et al. /Journal
of Affecti1.e Disorders 33 (I 995) 31-38 FQ: SUBSCORES AND TOTAL PHOBIA DIFFERENCES ON BASELINE
(28.6%) of the fluvoxamine group. The difference in the percentage of nonresponders is not significant. There were no significant differences between responders and nonresponders on sex, age or age at onset. Based on the assessment of blood plasma levels of brofaromine and fluvoxamine, it could be concluded that all patients were compliant to treatment. The differences in the scores between responders and nonresponders on the psychometric scales are depicted in Table 1. Responders had a baseline score on the HAS of 24.93 -t 2.57 and nonresponders had a score of 25.27 + 4.27. This difference was not significant. The difference of the scores on the HDS between responders (9.68 k 2.07) and nonresponders (10.80 f 2.11) is not significant. The FQ measures the amount of avoidance of a broad range of phobic stimuli on a 9-point scale ranging from 0 ‘would not avoid it’ to 8 ‘always avoid it’. The questionnaire consists of 15 items. The sum of the item scores constitutes the Total Phobia score. Three phobic subscores can also be calculated, each containing five items. The subscores are: the FQ-AG, the FQ-BI and the FQ-SP subscores. In Fig. 1, the FQ subscores and the Total Phobia score are depicted. Nonresponders showed significantly higher ratings on the FQ-BI subscore (F = 8.0; df = 1,41; P = 0.0072). The
Table 2 Fisher’s exact test (two-tail)
of indicator
variables
0
~~~-.--
FO-AG :
I __i___~
_~! ..A
I=0 BI
r--7
_ _..! TOTAL
FO-SP
RESPONDER
SCORE
NONRESPONDER n-15
“-28
Fig. 1. Baseline mean subscores and Total Phobia score of FQ (SEMI * Statistically significant difference between responders and nonresponders. TOTAL, Total Phobia score.
Total Phobia score just failed to reach statistical significance (P = 0.08). The indicator variables for the FQ are depicted in Table 2. There were no significant differences in the percentage of patients having a high score (2 20) on the FQ-AG or the FQ-SP. Of the nonresponders, 53.3% had a high score on the FQ-BI. 14.3% of the responders had a high score on the FQ-BI. This difference is significant (Fisher’s exact test; P = 0.0117). The indicator variables taking two subscores into account are depicted in Fig. 2. The differ-
of FQ P
Variable
Responder
Indicator variable: having a score > = 20 on FQ-AC Indicator variable: having a score > = 20 on FQ-SP Indicator variable: having a score > = 20 on FQ-BI Indicator variable: having a score >, = 20 on FQ-AC and on FQ-SP lndrcator variable: having a score >. = 20 on FQ-AC and on FQ-BI Indicator variable: having a score >, = 20 on FQ-SP and on FQ-BI Indicator variable: having a score >, = 20 on FQ-AC and on FQ-SP and on FQ-BI
85.7%
93.3%
NS
28.6%
40.0%
NS
14.3%
53.3%
P=O.O117
25.0%
33.3%
NS
10.7%
53.3%
P = 0.0040
0.0%
33.3%
P = 0.0031
0.0%
33.3%
P = 0.0031
to/c)
Nonresponder
t%)
B.R. Slaap et al. /Journal
of Affective Disorders 33 (I 995) 31-38
3.2. Prediction of nonresponse
PERCENTAGE OF PATIENTS WITH HIGH SCORES ON 2 FQ SUBSCORES (BASELINE)
1
* i
35
* -III
Fig. 2. Percentage of patients having a high score on two subscores of FQ simultaneously. * Statistically significant difference between responders and nonresponders. AG & SP > = 20: a score of 2 20 on both FQ-AG and FQ-SP subscores; AG & BI > = 20: a score of 2 20 on both FQ-AG and FQ-BI subscores; SP & RI > = 20: a score of t 20 OR both FQ-SP and FQ-BI subscores.
For the prediction of nonresponse to drug therapy, we carried out a logistic regression analysis on all significant baseline variables. Every single baseline variable significantly predicted nonresponse (P < 0.01). A best-fitting model was also calculated by carrying out a logistic regression analysis with all significant variables in the initial solution. The backwards elimination process resulted in a prediction based on one variable only. The indicator variable: ‘having a high score on the FQ-AG, the FQ-SP and on the FQ-BI’ was the only variable left in the equation. With this equation, the overall percentage correctly classified patients was 76.7%. All the responders were correctly classified but 10 nonresponders (66.7%) were incorrectly classified as responders. The fit of the model was reasonable ( x *=454df= 1.3 P=O.O330). -7 4. Discussion
ence in the percentage of patients with a high score on both the FQ-AG and the FQ-SP was not significant. Nonresponders more often had a high score on both the FQ-AG and the FQ-BI (Fisher’s exact test; P = 0.0040). There were no responders who had a high score on both the FQ-SP and on the FQ-BI. Of the nonresponders, 33.3% did have a high score on both these subscores. This result is significant (Fisher’s exact test; P = 0.0031). The same 33.3% of the nonresponders had a high score on all three subscores of the FQ (Fisher’s exact test; P = 0.0031). Two variables in Table 1 were not measured at baseline. These two variables from the UPI were measured in the 1st week of the study. Naturally, they can not be used as baseline predictors of nonresponse but they do give meaningful information on the number of panic attacks at the beginning of drug treatment. In week 1, nonresponders had almost twice as many panic attacks as responders (8.07 + 1.39 vs. 4.86 f 0.77). This difference is statistically significant (F = 4,87; df = 1,41; P = 0.0329). Nonresponders also reported to experience more tingling sensations (Kruskal-Wallis; P = 0.0199)
In this study, we found that nonresponders to drug therapy had a higher score on the FQ-BI subscore. A larger percentage of patients in the nonresponder group had a high score (of 2 20) on various FQ subscores. At the end of week 1, they had almost twice as many panic attacks. We were unable to replicate the findings of others that there are baseline differences between responders and nonresponders on the HAS (Liebowitz et al., 1986; Rosenberg et al., 1991) or the HDS (Rosenberg et al., 19911. A possible explanation for this might be that, in comparison with others, our sample was more homogeneous where HAS or HDS are concerned. The 95% confidence interval of the HAS in our sample (n = 44) was 19.08-31.92. The sample (n = 29) in the study of Liebowitz et al. (1986) had a 95% confidence interval of the HAS of 3.6-29.6. In the study of Rosenberg et al. (1991), the mean HAS and HDS of the whole sample (n = 123) was not stated but they did report on the mean ( f SD) of the responders and the nonresponders on these scales. The SD for the HAS and the HDS scores of responders and nonresponders was X5-6.8. In our sample, this value ranges
?h
B.R. Slaap et al. /Journal
of Affectir$e Disorders 33 (1995) 31-38
from 2.2 to 4.3. The scores on the HAS and the HDS in our sample had a smaller range which may have obscured differences (if any) between responders and nonresponders to become evident. Rosenberg et al. (1991) concluded that nonresponse was associated with more severe symptoms at baseline. Nonresponders in their study had higher scores on the HAS and the HDS and they had a lower panic frequency at baseline. In this study, we did not have a baseline panic frequency but our finding that nonresponders had a higher number of panic attacks at the end of week 1 seems to corroborate the conclusion of Rosenberg et al. (1991) that nonresponders are more severely ill. Woodman et al. (1994) found lower baseline phobic symptom level to be a predictor of response to alprazolam. They used the Overall Phobia Rating (Ballenger et al., 1988) to measure general phobic anxiety. This 11-point phobic anxiety scale ranges from 0 ‘no phobia’ to 10 ‘extremely distressing or restricting’. In our study, we used the FQ and its subscores which give more specific information on the phobic symptoms that the patients experience. The nonresponders in our sample had a higher score on the FQ-BI subscore and a greater percentage of them had a high score on several subscores. Our results seem to corroborate their findings since the Overall Phobia Rating measures distress caused by all phobias. In contrast to the findings of Woodman et al. (1994) that a high level of phobic anxiety is associated with poor treatment response, Maier et al. (1991) reported that patients with extensive avoidance behaviour had the most profit from the active drugs (alprazolam or imipramine). In their study, the effect of the treatment was most pronounced with respect to avoidance behaviour. In our study there were no statistically significant differences, as measured by the FQ-AG, between responders and nonresponders on agoraphobic behaviour. Comparison between studies is, however, hampered by the fact that Maier et al. (1991) and Woodman et al. (1994) used drugs with sedative properties whereas in our study we did not. It appears that the relationship between
agoraphobic behaviour and response to psychopharmacological treatment is unclear. Further research is needed to clarify what the influence is of the severity of agoraphobic avoidance on treatment response to antidepressant drugs. The higher scores of the nonresponders on the FQ-BI subscore and the larger percentage of patients with a high score on the FQ subscores might suggest that these nonresponders had comorbid phobic symptoms. It appeared that several patients had symptoms of a blood-injury phobia and/or a social phobia. Eight patients (53.3%) in the nonresponder group had a score of 2 20 on the FQ-BI compared with four patients (14.3%) in the responder group. Five patients (33.3%) in the nonresponder group had a score of 2 20 on both the FQ-BI and the FQ-SP. There were no such patients in the responder group. These patients did not fulfil diagnostic criteria for another (comorbid) anxiety disorder since this was one of the exclusion criteria. It is possible that some patients may have suffered from symptoms reminiscent of a simple phobia (like bloodinjury phobia) or a social phobia. The available literature on comorbidity among the anxiety disorders suggests that a large percentage of patients with PD has an additional diagnosis. The percentage of PD patients with an additional Axis I diagnosis ranges between 44% (Di Nardo and Barlow, 1990) and 83.3% (Starcevic et al., 1992). Differences in the reported percentages reflect differences in diagnostic criteria and differences between samples. It should be emphasized that the degree of diagnostic comorbidity is directly related to the threshold set to determine the presence or absence of various disorders (Frances et al., 1990). In the above-cited studies as well as in other papers (Brown and Barlow, 1992; de Ruiter et al., 1989; Sanderson et al., 19901, it appears that simple phobia or social phobia is the most often occurring additional diagnosis in PD. The percentages for a comorbid simple phobia vary between 6.3% (Brown and Barlow, 1992) and 46.6% (de Ruiter et al., 1989). For a comorbid social phobia, the percentages vary between 9.6% (de Ruiter et al., 1989) and 40.7% (Starcevic et al., 19921 (all percentages are summed over all PD patients).
B.R. Slaap et al. /Journal
of Affective Disorders 33 (1995) 31-38
If we compare our findings of baseline phobic symptoms with the findings in the literature on cotnorbidity, it appears that our results are comparable. Of the 43 patients in the study, 12 patients (27.9%) had a high score on the FQ-BI. There are 14 patients (32.6%) with a high score on the FQ-SP. Both these percentages fall in the ranges cited above. We have found that nonresponders to drug treatment more often have comorbid phobic symptoms as compared with responders. A similar finding has been reported before: Pollack et al. (1993) found in their follow-up study that a comorbid social phobia was associated with poor treatment response. That a blood-injury phobia can interfere with the treatment of PD has been noted before. Di Nardo and Barlow (1990) described a case of an agoraphobic patient with a distinct blood-injury phobia. This phobia interfered with their treatment programme in such a way that it needed to be treated first. Emmelkamp and Bouman (1991) discussed specific difficulties and complications in the treatment of PD. Among the additional anxiety disorders that complicate the treatment of PD, they described blood-injury phobia as well. Further research is needed to find out what the influence is of comorbid anxiety symptoms or syndromes on the treatment of PD. The second aim of the study was the prediction of nonresponse to drug therapy. We first predicted nonresponse by using every significant baseline scale separately. All these variables significantly predicted treatment outcome. In addition, we aimed to create a best-fitting model by entering all significant baseline variables in a stepwise logistic regression analysis. This resulted in a prediction based on one variable only. The indicator variable ‘having a high score on all three FQ subscores’ was the only variable in the model. The other indicator variables were subsets of the indicator variable in the equation, so they did not add to prediction. The FQ-BI also did not add significantly to the prediction to be entered in the equation by the logistic regression routine. We had hoped to base a prediction on several variables but this was impossible because not that many variables were significant at baseline.
37
In this preliminary report, we have found that on baseline nonresponders to drug therapy differ from responders in that they have a higher FQ-BI subscore. They more often had a high score on FQ subscores, indicative of comorbid phobic symptoms. This predicted nonresponse to drug treatment. A methodological flaw of the present study might be the limited number of patients that was studied. In future studies, we intend to include larger numbers of patients, possibly enabling us to further substantiate present findings.
References Albus, M., Lecrubier, Y., Maier, W., Buller, R., Rosenberg, R. and Hippius, H. (1990) Drug treatment of panic disorder: early response to treatment as a predictor of final outcome. Acta Psychiatr. Stand. 82, 359-365. American Psychiatric Association (1987) DSM-III-R: Diagnostic and Statistical Manual of Mental Disorders. 3rd Rev. Ed. American Psychiatric Association, Washington, DC. Ballenger, J.C., Burrows, G.D., Du Pont, R.L., Lesser, I.M., Noyes, R., Pecknold, J., Rifkin, A. and Swinson, R. (1988) Alprazolam in panic disorder and agoraphobia: results of a multi-center trial. I. Efficacy in short term treatment. Arch. Gen. Psychiatry 45, 413-422. Brown, T.A. and Barlow, D.H. (1992) Comorbidity among anxiety disorders: implications for treatment and DSM-IV. J. Cons. Clin. Psychol. 60, 835-844. den Boer, J.A. and Wesrenberg, H. (i990) Do panic attacks reflect an abnormality in serotonin receptor subtypes? Human Psychopharmacol. 6, 25-30. de Ruiter, C., Rijken, H., Garssen, B., van Schaik, A. and Kraaimaat, F. (1989) Comorbidity among the anxiety disorders. J. Anx. Dis. 3, 57-68. Di Nardo, P.A. and Barlow, D.H. (1990) Syndrome and symptom comorbidity in the anxiety disorders. In: J.D. Maser and C.R. Cloninger (Eds.), Comorbidity in Anxiety and Depression. American Psychiatric Press, Washington, DC, pp. 205-230. Emmelkamp, P.M.G. and Bouman, T.K. (1991) Psychological approaches to the difficult patient. In: J.R. Walker et al. (Eds.), Panic Disorder and Agoraphobia: A Comprehensive Guide for The Practitioner. Brooks-Cole, Pacific Grove, CA, pp, 398-429. Emmelkamp, P.M.G. and van den Hout, A. (1983) Failure in treating agoraphobia. In: E.B. Foa and P.M.G. Emmelkamp (Eds.), Failures in Behaviour Therapy. Wiley, New York, NY, pp. 58-81. Emmelkamp, P.M.G. and Foa, E.B. (1983) Failures are a challenge. In: E.B. Foa and P.M.G. Emmelkamp (Eds.), Failures in Behaviour Therapy. Wiley, New York, NY, pp. l-10.
3x
B. R. Slaap et al. /Journal
of Affectke Disorders 33 (1995) 31-38
Frances, A., Widiger, T. and Fyer, M.R. (19901 The influence of classification methods on comorbidity. In: J.D. Maser and C.R. Cloninger (Eds.), Comorbidity in Anxiety and Depression. American Psychiatric Press, Washington, DC, pp. 41-59. Hamilton, M. (1959) The assessment of anxiety states by rating. Br. J. Med. Psychol. 32, 50-55. Hamilton, M. (1967) Development of a rating scale for primary depressive illness. Br. J. Sot. Clin. Psychol. 6. 278296. Joyce, P.R. and Paykel, E.S. (1989) Predictors of drug response in depression. Arch. Gen. Psychiatry 46, 89-99. Liebowitz, M.R., Fyer, A.J., Gorman, J.M., Campeas, R., Levin. A., Davies, S.R., Goetz, D. and Klein, D.F. (1986) Alprazolam in the treatment of panic disorders. J. Clin. Psychopharmacol. 6. 13-20. Liebowitz. M.R. (1989) Antidepressants in panic disorders. Br. J. Psychiatry 155 (Suppl. 61, 46-52. Maier, W., Roth, M., Buller, R., Argyle, N., Rosenberg, R., Branson. S. and Benkert, 0. (1991) Agoraphobia in panic disorder: an indicator of the severity of panic disorder or a distinct entity? Psychiatr. Ann. 21, 374-381. Marks, I.M. and Mathews, A.M. (1979) Brief standard selfrating for phobic patients. Behav. Res. Ther. 17, 263-267. Modigh, K. (1987) Antidepressant drugs in anxiety disorders. Acta Psychiatr. Stand. 76 (Suppl. 3351, 57-71. Oei, T.P.S., Moylan, A. and Evans, L. (1991) Validity and clinical utility of the Fear Questionnaire for anxiety-disorder patients. Psychol. Assessment 3, 391-397. Pollack, M.H., Otto, M.W., Tesar, G.E., Cohen, L.S., Meltzer-Brody, S. and Rosenbaum, J.F. (1993) Long-term outcome after acute treatment with alprazolam or clonazepam for panic disorder. J. Clin. Psychopharmacol. 13, 257-263.
Rosenberg, R., Beth, P.. Mellergird, M. and Ottosson, J-O. (1991) Alprazolam, imipramine and placebo treatment of panic disorder: predicting therapeutic response. Acta Psychiatr. Stand. Suppl. 365, 46-52. Sanderson, W.C., Di Nardo, P.A., Rapee, R.M. and Barlow. D.H. (1990) Syndrome comorbidity in patients diagnosed with a DSM-III-R anxiety disorder. J. Abnorm. Psychol. 99, 308-312. Scheibe. G., Nutzinger, D., Buller, R. and Walter, A.-U. (1992) Pretreatment anxiety level as differential predictor in outpatients with panic disorder. Arzneim. Forsch./Drug Res. 42. 1090-1094. Starcevic, V., Uhlenhuth, E.H., Kellner, R. and Pathak, D., (1992) Patterns of comorbidity in panic disorder and agoraphobia. Psychiatr. Res. 42, 171-183. Tyrer, P. and Shawcross, C. (1988) Monoamine oxidase inhibitors in anxiety disorders. J. Psychiatr. Res. 22. 87-98. van Vliet, I.M., Westenberg, H. and den Boer, J.A. (1993) MAO-inhibitors in Panic Disorder: clinical effects of treatment with brofaromine, a double blind placebo controlled study. Psychopharmacology, 112, 483-489. van Vliet, I.M., Slaap, B.R., den Boer, J.A. and Westenberg, H. (1994) A comparison of brofaromine and fluvoxamine in panic disorder, a double blind study. Submitted for publication. Westenberg, H. and den Boer, J.A. (19941 The neuropharmacology of anxiety; a review on the role of serotonin. In: J.A. den Boer and J.M.A. Sitsen (Eds.1, Handbook of Depression and Anxiety. Marcel Dekker. New York, NY, pp. 405-446. Woodman, C.L., Noyes, Jr., R., Ballenger, J.C. Lydiard, R.B., Sievers. G. and Mihalco, D. (1994) Predictors of response to alprazolam and placebo in patients with panic disorder. .I. Affect. Disord. 30, 5-13.
OF
AFFECTIVE DISORDERS
EUEVIER
Journal
of Affective
Disorders
33 (1995) 31-38
Phobic symptoms as predictors of nonresponse to drug therapy in panic disorder patients (a preliminary report) Bernhard R. Slaap *, Irene M. van Vliet, Herman G.M. Westenberg, Johan A. den Boer Department of Psychiatry,
Rudolf Magnus Institute for Neuroscience, Academic Hospital Utrecht, PO Box 85500, 3508 GA Utrecht, The Netherlands Received
4 May 1994; Accepted
5 September
1994
Abstract that predict nonresponse to drug therapy (brofaromine or fluvoxamine) were investigated in a sample of patients. We used a strict definition of nonresponse to find patients who did not respond at all after 12 weeks of treatment. Using this definition, 15 patients (32.6%) were considered nonresponders. Nonresponders had a higher score on the Blood-Injury subscore of the Fear Questionnaire (FQ) and more often had high scores on several FQ subscores, indicative of comorbid phobic symptoms. These variables were subsequently used to Factors
44 panic disorder
predict nonresponse. Keywords:
Panic disorder;
Prediction
of response;
Brofaromine;
1. Introduction
The principal management strategy for patients with panic disorder (PD) with or without agoraphobia (DSM-III-R; American Psychiatric Association, 1987) is drug therapy in combination with behaviour therapy. Various drugs have been shown to be effective in the treatment of PD: tricyclic antidepressants (for review, see Modigh, 1987, and Liebowitz, 1989), selective serotonin reuptake inhibitors (SSRIs; for review, see den Boer and Westenberg, 1990, and Westenberg and den Boer, 1994) and monoamine oxidase in-
* Corresponding
author.
Fax: (31) (30) 541844.
01650327/95/$09.50 0 1995 Elsevier SSDI 0165-0327(94)00070-O
Science
Fluvoxamine
hibitors (MAOIS; for review, see Tyrer and Shawcross, 1988). In all these studies, beneficial effects on the number of panic attacks were reported. However, not all patients respond to treatment: 20-40% of the patients are nonresponders. At present, it is unclear whether there are differences in psychological or biological characteristics of responders vs. nonresponders. Predicting response or nonresponse to antidepressant drugs would be very helpful since it takes - 4 weeks for these drugs to become clinically effective. There is scarce literature on predictors of response in PD. Compared with the vast amount of literature on the prediction of response in depression (see for, e.g., Joyce and Paykel, 19891, this area still seems relatively unex-
B.V. All rights reserved
32
B.R. Sap
et al. /Journal
of Affectit,e Disorder.\ 33 (1995) 31-38
plored in PD. Some predictors of response to drug therapy have been reported in the literature: (1) early improvement on the number of spontaneous panic attacks during treatment with alprazolam (Albus et al., 1990); (2) lower baseline scores on the Hamilton Anxiety Scale (HAS; Hamilton, 1959) (Liebowitz et al., 1986; Rosenberg et al., 1991; Scheibe et al., 1992; Woodman et al., 1994); (3) lower baseline scores on the Hamilton Depression Scale (HDS; Hamilton, 1967) (Rosenberg et al., 1991); (4) age > 40 (Woodman et al., 1994); and (5) lower baseline levels of phobic symptoms (Woodman et al., 1994). In the present study, we report on possible predictors for nonresponse to drug treatment in PD patients. The idea to study the nonresponders is not without precedent: there have been others who have tried to define clinical characteristics of nonresponders to, e.g., behaviour therapy (Emmelkamp and Foa, 1983; Emmelkamp and van den Hout, 1983). The aim of the present study is to look for clinical variables, measured at baseline, which may distinguish responders from nonresponders in a patient sample suffering from PD. The variables that differ significantly between the groups will be used to predict nonresponsc to drug therapy.
2. Methods 2 1. Design
For the present study, the data of two previously published studies were pooled. In these studies brofaromine, a selective and reversible MAOI, was compared with either fluvoxamine, an SSRI (van Vliet et al., 1994), or placebo (van Vliet et al., 1993). Both studies had a double-blind design and lasted 12 weeks. The psychometric scales were similar in both studies. By pooling the data, we obtained a data base of 60 patients. 29 patients were treated with brofaromine, 15 patients were treated with fluvoxamine and 16 patients received placebo. The 44 patients who received drug treatment have been studied in this paper. The patients treated with placebo were excluded from the study because we were inter-
ested predictors of nonresponse to active drug treatment. Possible predictors of nonresponse to placebo will be the subject of another article. The patients who received drug treatment were analysed as one group in this study. ANOVA with drug (brofaromine or fluvoxamine) as covariate showed no significant differences between the drug groups. Furthermore, from the study of van Vliet et al. (1994) it appears that there are no significant differences between brofaromine and fluvoxamine on any of the outcome measures. Nonresponders were defined as patients who still had panic attacks at endpoint and who had a reduction of < 50% on the Agoraphobia subscore of the Fear Questionnaire (FQ; Marks and Mathews. 1979). Patients who were free of panic attacks at endpoint and/or showed an improvement of 2 50% on the Agoraphobia subscore were considered to be responders. 2.2. Patients For each study, 30 patients were recruited from the Outpatient Clinic, Department of Biological Psychiatry, University Hospital, Utrecht, The Netherlands. The studies were approved by the Ethics Committee of the Academic Hospital and informed consent was obtained from all patients. Included in the studies were patients suffering from PD with Agoraphobia (59 patients) or without Agoraphobia (one patient) according to DSM-III-R criteria. Excluded were patients with other anxiety disorders, major affective disorder or psychotic disorder, alcohol abuse, severe sleep disturbance and patients suffering from other medical problems on the basis of a complete medical evaluation. Patients with a score of r 15 on the HDS were also excluded. During treatment, the use of other psychotropic drugs was not allowed except the use of oxazepam to a maximum of 20 mg daily if required. 60 patients, 54 females and six males were enrolled and treated for 12 weeks using a double-blind placebo controlled design. Mean age (*SD) was 36.6 f 7.1 years (range 20-50) and the mean duration of illness was 9.7 k 6.2 years (range 0.6-32). In each study, patients were randomly allocated to one of the two treatment groups: bro-
B.R. Slaap et al. /Journal
of Affective Disorders 33 (1995) 31-38
faromine or placebo in one study (van Vliet et al., 1993), or brofaromine or fluvoxamine in the other (van Vliet et al., 1994). 2.3. Symptom assessment At baseline and at the end of treatment, the HDS was completed. Efficacy of treatment was further assessed using the FQ and the HAS. The FQ and the HAS were completed on baseline and at the end of weeks 1, 2, 4 and 8 and at the end of week 12. The Utrecht Panic Inventory (UPI) was completed daily. This is a questionnaire, based on the DSM-III-R definition of a panic attack, measuring the frequency and intensity of panic attacks and the associated symptoms. Blood plasma levels of brofaromine or fluvoxamine were monitored to ensure compliance to treatment.
33
were computed which divided patients in high or low scorers on a FQ subscore. Patients were considered high scorers on a FQ subscore when they had a score of 2 20. (The maximum score on subscore is 40.) From normative data of the FQ in the article of Oei et al. (19911, it appears that approximately half of the patients with PD with agoraphobia had a score of 2 20 on the FQ Agoraphobia subscore (FQ-AG). N 30% of the patients had a score of 2 20 on the FQ Social Phobia subscore (FQ-SP) and N 25% had a score of 2 20 on the FQ Blood-Injury phobia subscore (FQ-BI). For the prediction of nonresponse to drug therapy, a logistic regression analysis was carried out for all significant baseline variables. Subsequently, a best-fitting model was made with all significant variables using logistic regression with backwards elimination. As criterion for variable selection, the likelihood-ratio test was used.
2.4. Data analysis Data were analysed using a commercially available statistical package (SPSS). To analyse the differences between responders and nonresponders, an ANOVA was performed on the baseline variables. When the assumption of nora nonparametric test mality was violated, (Kruskal-Wallis ANOVA) was applied. The categorical data were tested with the Fisher’s exact test (two-tail). To further investigate the influence of phobic symptoms on nonresponse, indicator variables
Table 1 ANOVA/Kruskal-Wallis
of psychometric
3. Results 3.1. Clinical variables
Of the 44 patients receiving medication (brofaromine or fluvoxaminel, there was one patient who did not complete the UPI, therefore, this patient was not entered into the statistical analysis. Of the remaining 43 patients, 15 patients (32.6%) were nonresponders, 11 patients (37.9%) of the brofaromine group and four patients
scales
Variable
Responder
HAS HDS FQ Total Score FQ-AG subscore FQ-SP subscore FQ-BI subscore
24.93 9.68 56.75 30.36 14.79 11.61
? + k f f +
(mean + SEMI
2.57 2.07 3.09 8.13 9.26 1.44
Nonresponder 25.27 10.80 66.07 30.67 17.53 17.87
+ + + + + k
4.27 2.11 4.25 7.35 8.54 1.37
UP1 week 1: number
of panic attacks
4.86 + 0.77
8.07 ? 1.39
UPI week 1: tingling
sensations
0.64 f 0.17
1.94 f 0.49
(mean f SEM)
P NS NS NS NS NS F = 8.00;df = 1,41; P = 0.0072 F = 4,87; df = 1.41: P = 0.0329 Kruskal-Wallis; P = 0.0199
34
B. R. Slaap et al. /Journal
of Affecti1.e Disorders 33 (I 995) 31-38 FQ: SUBSCORES AND TOTAL PHOBIA DIFFERENCES ON BASELINE
(28.6%) of the fluvoxamine group. The difference in the percentage of nonresponders is not significant. There were no significant differences between responders and nonresponders on sex, age or age at onset. Based on the assessment of blood plasma levels of brofaromine and fluvoxamine, it could be concluded that all patients were compliant to treatment. The differences in the scores between responders and nonresponders on the psychometric scales are depicted in Table 1. Responders had a baseline score on the HAS of 24.93 -t 2.57 and nonresponders had a score of 25.27 + 4.27. This difference was not significant. The difference of the scores on the HDS between responders (9.68 k 2.07) and nonresponders (10.80 f 2.11) is not significant. The FQ measures the amount of avoidance of a broad range of phobic stimuli on a 9-point scale ranging from 0 ‘would not avoid it’ to 8 ‘always avoid it’. The questionnaire consists of 15 items. The sum of the item scores constitutes the Total Phobia score. Three phobic subscores can also be calculated, each containing five items. The subscores are: the FQ-AG, the FQ-BI and the FQ-SP subscores. In Fig. 1, the FQ subscores and the Total Phobia score are depicted. Nonresponders showed significantly higher ratings on the FQ-BI subscore (F = 8.0; df = 1,41; P = 0.0072). The
Table 2 Fisher’s exact test (two-tail)
of indicator
variables
0
~~~-.--
FO-AG :
I __i___~
_~! ..A
I=0 BI
r--7
_ _..! TOTAL
FO-SP
RESPONDER
SCORE
NONRESPONDER n-15
“-28
Fig. 1. Baseline mean subscores and Total Phobia score of FQ (SEMI * Statistically significant difference between responders and nonresponders. TOTAL, Total Phobia score.
Total Phobia score just failed to reach statistical significance (P = 0.08). The indicator variables for the FQ are depicted in Table 2. There were no significant differences in the percentage of patients having a high score (2 20) on the FQ-AG or the FQ-SP. Of the nonresponders, 53.3% had a high score on the FQ-BI. 14.3% of the responders had a high score on the FQ-BI. This difference is significant (Fisher’s exact test; P = 0.0117). The indicator variables taking two subscores into account are depicted in Fig. 2. The differ-
of FQ P
Variable
Responder
Indicator variable: having a score > = 20 on FQ-AC Indicator variable: having a score > = 20 on FQ-SP Indicator variable: having a score > = 20 on FQ-BI Indicator variable: having a score >, = 20 on FQ-AC and on FQ-SP lndrcator variable: having a score >. = 20 on FQ-AC and on FQ-BI Indicator variable: having a score >, = 20 on FQ-SP and on FQ-BI Indicator variable: having a score >, = 20 on FQ-AC and on FQ-SP and on FQ-BI
85.7%
93.3%
NS
28.6%
40.0%
NS
14.3%
53.3%
P=O.O117
25.0%
33.3%
NS
10.7%
53.3%
P = 0.0040
0.0%
33.3%
P = 0.0031
0.0%
33.3%
P = 0.0031
to/c)
Nonresponder
t%)
B.R. Slaap et al. /Journal
of Affective Disorders 33 (I 995) 31-38
3.2. Prediction of nonresponse
PERCENTAGE OF PATIENTS WITH HIGH SCORES ON 2 FQ SUBSCORES (BASELINE)
1
* i
35
* -III
Fig. 2. Percentage of patients having a high score on two subscores of FQ simultaneously. * Statistically significant difference between responders and nonresponders. AG & SP > = 20: a score of 2 20 on both FQ-AG and FQ-SP subscores; AG & BI > = 20: a score of 2 20 on both FQ-AG and FQ-BI subscores; SP & RI > = 20: a score of t 20 OR both FQ-SP and FQ-BI subscores.
For the prediction of nonresponse to drug therapy, we carried out a logistic regression analysis on all significant baseline variables. Every single baseline variable significantly predicted nonresponse (P < 0.01). A best-fitting model was also calculated by carrying out a logistic regression analysis with all significant variables in the initial solution. The backwards elimination process resulted in a prediction based on one variable only. The indicator variable: ‘having a high score on the FQ-AG, the FQ-SP and on the FQ-BI’ was the only variable left in the equation. With this equation, the overall percentage correctly classified patients was 76.7%. All the responders were correctly classified but 10 nonresponders (66.7%) were incorrectly classified as responders. The fit of the model was reasonable ( x *=454df= 1.3 P=O.O330). -7 4. Discussion
ence in the percentage of patients with a high score on both the FQ-AG and the FQ-SP was not significant. Nonresponders more often had a high score on both the FQ-AG and the FQ-BI (Fisher’s exact test; P = 0.0040). There were no responders who had a high score on both the FQ-SP and on the FQ-BI. Of the nonresponders, 33.3% did have a high score on both these subscores. This result is significant (Fisher’s exact test; P = 0.0031). The same 33.3% of the nonresponders had a high score on all three subscores of the FQ (Fisher’s exact test; P = 0.0031). Two variables in Table 1 were not measured at baseline. These two variables from the UPI were measured in the 1st week of the study. Naturally, they can not be used as baseline predictors of nonresponse but they do give meaningful information on the number of panic attacks at the beginning of drug treatment. In week 1, nonresponders had almost twice as many panic attacks as responders (8.07 + 1.39 vs. 4.86 f 0.77). This difference is statistically significant (F = 4,87; df = 1,41; P = 0.0329). Nonresponders also reported to experience more tingling sensations (Kruskal-Wallis; P = 0.0199)
In this study, we found that nonresponders to drug therapy had a higher score on the FQ-BI subscore. A larger percentage of patients in the nonresponder group had a high score (of 2 20) on various FQ subscores. At the end of week 1, they had almost twice as many panic attacks. We were unable to replicate the findings of others that there are baseline differences between responders and nonresponders on the HAS (Liebowitz et al., 1986; Rosenberg et al., 1991) or the HDS (Rosenberg et al., 19911. A possible explanation for this might be that, in comparison with others, our sample was more homogeneous where HAS or HDS are concerned. The 95% confidence interval of the HAS in our sample (n = 44) was 19.08-31.92. The sample (n = 29) in the study of Liebowitz et al. (1986) had a 95% confidence interval of the HAS of 3.6-29.6. In the study of Rosenberg et al. (1991), the mean HAS and HDS of the whole sample (n = 123) was not stated but they did report on the mean ( f SD) of the responders and the nonresponders on these scales. The SD for the HAS and the HDS scores of responders and nonresponders was X5-6.8. In our sample, this value ranges
?h
B.R. Slaap et al. /Journal
of Affectir$e Disorders 33 (1995) 31-38
from 2.2 to 4.3. The scores on the HAS and the HDS in our sample had a smaller range which may have obscured differences (if any) between responders and nonresponders to become evident. Rosenberg et al. (1991) concluded that nonresponse was associated with more severe symptoms at baseline. Nonresponders in their study had higher scores on the HAS and the HDS and they had a lower panic frequency at baseline. In this study, we did not have a baseline panic frequency but our finding that nonresponders had a higher number of panic attacks at the end of week 1 seems to corroborate the conclusion of Rosenberg et al. (1991) that nonresponders are more severely ill. Woodman et al. (1994) found lower baseline phobic symptom level to be a predictor of response to alprazolam. They used the Overall Phobia Rating (Ballenger et al., 1988) to measure general phobic anxiety. This 11-point phobic anxiety scale ranges from 0 ‘no phobia’ to 10 ‘extremely distressing or restricting’. In our study, we used the FQ and its subscores which give more specific information on the phobic symptoms that the patients experience. The nonresponders in our sample had a higher score on the FQ-BI subscore and a greater percentage of them had a high score on several subscores. Our results seem to corroborate their findings since the Overall Phobia Rating measures distress caused by all phobias. In contrast to the findings of Woodman et al. (1994) that a high level of phobic anxiety is associated with poor treatment response, Maier et al. (1991) reported that patients with extensive avoidance behaviour had the most profit from the active drugs (alprazolam or imipramine). In their study, the effect of the treatment was most pronounced with respect to avoidance behaviour. In our study there were no statistically significant differences, as measured by the FQ-AG, between responders and nonresponders on agoraphobic behaviour. Comparison between studies is, however, hampered by the fact that Maier et al. (1991) and Woodman et al. (1994) used drugs with sedative properties whereas in our study we did not. It appears that the relationship between
agoraphobic behaviour and response to psychopharmacological treatment is unclear. Further research is needed to clarify what the influence is of the severity of agoraphobic avoidance on treatment response to antidepressant drugs. The higher scores of the nonresponders on the FQ-BI subscore and the larger percentage of patients with a high score on the FQ subscores might suggest that these nonresponders had comorbid phobic symptoms. It appeared that several patients had symptoms of a blood-injury phobia and/or a social phobia. Eight patients (53.3%) in the nonresponder group had a score of 2 20 on the FQ-BI compared with four patients (14.3%) in the responder group. Five patients (33.3%) in the nonresponder group had a score of 2 20 on both the FQ-BI and the FQ-SP. There were no such patients in the responder group. These patients did not fulfil diagnostic criteria for another (comorbid) anxiety disorder since this was one of the exclusion criteria. It is possible that some patients may have suffered from symptoms reminiscent of a simple phobia (like bloodinjury phobia) or a social phobia. The available literature on comorbidity among the anxiety disorders suggests that a large percentage of patients with PD has an additional diagnosis. The percentage of PD patients with an additional Axis I diagnosis ranges between 44% (Di Nardo and Barlow, 1990) and 83.3% (Starcevic et al., 1992). Differences in the reported percentages reflect differences in diagnostic criteria and differences between samples. It should be emphasized that the degree of diagnostic comorbidity is directly related to the threshold set to determine the presence or absence of various disorders (Frances et al., 1990). In the above-cited studies as well as in other papers (Brown and Barlow, 1992; de Ruiter et al., 1989; Sanderson et al., 19901, it appears that simple phobia or social phobia is the most often occurring additional diagnosis in PD. The percentages for a comorbid simple phobia vary between 6.3% (Brown and Barlow, 1992) and 46.6% (de Ruiter et al., 1989). For a comorbid social phobia, the percentages vary between 9.6% (de Ruiter et al., 1989) and 40.7% (Starcevic et al., 19921 (all percentages are summed over all PD patients).
B.R. Slaap et al. /Journal
of Affective Disorders 33 (1995) 31-38
If we compare our findings of baseline phobic symptoms with the findings in the literature on cotnorbidity, it appears that our results are comparable. Of the 43 patients in the study, 12 patients (27.9%) had a high score on the FQ-BI. There are 14 patients (32.6%) with a high score on the FQ-SP. Both these percentages fall in the ranges cited above. We have found that nonresponders to drug treatment more often have comorbid phobic symptoms as compared with responders. A similar finding has been reported before: Pollack et al. (1993) found in their follow-up study that a comorbid social phobia was associated with poor treatment response. That a blood-injury phobia can interfere with the treatment of PD has been noted before. Di Nardo and Barlow (1990) described a case of an agoraphobic patient with a distinct blood-injury phobia. This phobia interfered with their treatment programme in such a way that it needed to be treated first. Emmelkamp and Bouman (1991) discussed specific difficulties and complications in the treatment of PD. Among the additional anxiety disorders that complicate the treatment of PD, they described blood-injury phobia as well. Further research is needed to find out what the influence is of comorbid anxiety symptoms or syndromes on the treatment of PD. The second aim of the study was the prediction of nonresponse to drug therapy. We first predicted nonresponse by using every significant baseline scale separately. All these variables significantly predicted treatment outcome. In addition, we aimed to create a best-fitting model by entering all significant baseline variables in a stepwise logistic regression analysis. This resulted in a prediction based on one variable only. The indicator variable ‘having a high score on all three FQ subscores’ was the only variable in the model. The other indicator variables were subsets of the indicator variable in the equation, so they did not add to prediction. The FQ-BI also did not add significantly to the prediction to be entered in the equation by the logistic regression routine. We had hoped to base a prediction on several variables but this was impossible because not that many variables were significant at baseline.
37
In this preliminary report, we have found that on baseline nonresponders to drug therapy differ from responders in that they have a higher FQ-BI subscore. They more often had a high score on FQ subscores, indicative of comorbid phobic symptoms. This predicted nonresponse to drug treatment. A methodological flaw of the present study might be the limited number of patients that was studied. In future studies, we intend to include larger numbers of patients, possibly enabling us to further substantiate present findings.
References Albus, M., Lecrubier, Y., Maier, W., Buller, R., Rosenberg, R. and Hippius, H. (1990) Drug treatment of panic disorder: early response to treatment as a predictor of final outcome. Acta Psychiatr. Stand. 82, 359-365. American Psychiatric Association (1987) DSM-III-R: Diagnostic and Statistical Manual of Mental Disorders. 3rd Rev. Ed. American Psychiatric Association, Washington, DC. Ballenger, J.C., Burrows, G.D., Du Pont, R.L., Lesser, I.M., Noyes, R., Pecknold, J., Rifkin, A. and Swinson, R. (1988) Alprazolam in panic disorder and agoraphobia: results of a multi-center trial. I. Efficacy in short term treatment. Arch. Gen. Psychiatry 45, 413-422. Brown, T.A. and Barlow, D.H. (1992) Comorbidity among anxiety disorders: implications for treatment and DSM-IV. J. Cons. Clin. Psychol. 60, 835-844. den Boer, J.A. and Wesrenberg, H. (i990) Do panic attacks reflect an abnormality in serotonin receptor subtypes? Human Psychopharmacol. 6, 25-30. de Ruiter, C., Rijken, H., Garssen, B., van Schaik, A. and Kraaimaat, F. (1989) Comorbidity among the anxiety disorders. J. Anx. Dis. 3, 57-68. Di Nardo, P.A. and Barlow, D.H. (1990) Syndrome and symptom comorbidity in the anxiety disorders. In: J.D. Maser and C.R. Cloninger (Eds.), Comorbidity in Anxiety and Depression. American Psychiatric Press, Washington, DC, pp. 205-230. Emmelkamp, P.M.G. and Bouman, T.K. (1991) Psychological approaches to the difficult patient. In: J.R. Walker et al. (Eds.), Panic Disorder and Agoraphobia: A Comprehensive Guide for The Practitioner. Brooks-Cole, Pacific Grove, CA, pp, 398-429. Emmelkamp, P.M.G. and van den Hout, A. (1983) Failure in treating agoraphobia. In: E.B. Foa and P.M.G. Emmelkamp (Eds.), Failures in Behaviour Therapy. Wiley, New York, NY, pp. 58-81. Emmelkamp, P.M.G. and Foa, E.B. (1983) Failures are a challenge. In: E.B. Foa and P.M.G. Emmelkamp (Eds.), Failures in Behaviour Therapy. Wiley, New York, NY, pp. l-10.
3x
B. R. Slaap et al. /Journal
of Affectke Disorders 33 (1995) 31-38
Frances, A., Widiger, T. and Fyer, M.R. (19901 The influence of classification methods on comorbidity. In: J.D. Maser and C.R. Cloninger (Eds.), Comorbidity in Anxiety and Depression. American Psychiatric Press, Washington, DC, pp. 41-59. Hamilton, M. (1959) The assessment of anxiety states by rating. Br. J. Med. Psychol. 32, 50-55. Hamilton, M. (1967) Development of a rating scale for primary depressive illness. Br. J. Sot. Clin. Psychol. 6. 278296. Joyce, P.R. and Paykel, E.S. (1989) Predictors of drug response in depression. Arch. Gen. Psychiatry 46, 89-99. Liebowitz, M.R., Fyer, A.J., Gorman, J.M., Campeas, R., Levin. A., Davies, S.R., Goetz, D. and Klein, D.F. (1986) Alprazolam in the treatment of panic disorders. J. Clin. Psychopharmacol. 6. 13-20. Liebowitz. M.R. (1989) Antidepressants in panic disorders. Br. J. Psychiatry 155 (Suppl. 61, 46-52. Maier, W., Roth, M., Buller, R., Argyle, N., Rosenberg, R., Branson. S. and Benkert, 0. (1991) Agoraphobia in panic disorder: an indicator of the severity of panic disorder or a distinct entity? Psychiatr. Ann. 21, 374-381. Marks, I.M. and Mathews, A.M. (1979) Brief standard selfrating for phobic patients. Behav. Res. Ther. 17, 263-267. Modigh, K. (1987) Antidepressant drugs in anxiety disorders. Acta Psychiatr. Stand. 76 (Suppl. 3351, 57-71. Oei, T.P.S., Moylan, A. and Evans, L. (1991) Validity and clinical utility of the Fear Questionnaire for anxiety-disorder patients. Psychol. Assessment 3, 391-397. Pollack, M.H., Otto, M.W., Tesar, G.E., Cohen, L.S., Meltzer-Brody, S. and Rosenbaum, J.F. (1993) Long-term outcome after acute treatment with alprazolam or clonazepam for panic disorder. J. Clin. Psychopharmacol. 13, 257-263.
Rosenberg, R., Beth, P.. Mellergird, M. and Ottosson, J-O. (1991) Alprazolam, imipramine and placebo treatment of panic disorder: predicting therapeutic response. Acta Psychiatr. Stand. Suppl. 365, 46-52. Sanderson, W.C., Di Nardo, P.A., Rapee, R.M. and Barlow. D.H. (1990) Syndrome comorbidity in patients diagnosed with a DSM-III-R anxiety disorder. J. Abnorm. Psychol. 99, 308-312. Scheibe. G., Nutzinger, D., Buller, R. and Walter, A.-U. (1992) Pretreatment anxiety level as differential predictor in outpatients with panic disorder. Arzneim. Forsch./Drug Res. 42. 1090-1094. Starcevic, V., Uhlenhuth, E.H., Kellner, R. and Pathak, D., (1992) Patterns of comorbidity in panic disorder and agoraphobia. Psychiatr. Res. 42, 171-183. Tyrer, P. and Shawcross, C. (1988) Monoamine oxidase inhibitors in anxiety disorders. J. Psychiatr. Res. 22. 87-98. van Vliet, I.M., Westenberg, H. and den Boer, J.A. (1993) MAO-inhibitors in Panic Disorder: clinical effects of treatment with brofaromine, a double blind placebo controlled study. Psychopharmacology, 112, 483-489. van Vliet, I.M., Slaap, B.R., den Boer, J.A. and Westenberg, H. (1994) A comparison of brofaromine and fluvoxamine in panic disorder, a double blind study. Submitted for publication. Westenberg, H. and den Boer, J.A. (19941 The neuropharmacology of anxiety; a review on the role of serotonin. In: J.A. den Boer and J.M.A. Sitsen (Eds.1, Handbook of Depression and Anxiety. Marcel Dekker. New York, NY, pp. 405-446. Woodman, C.L., Noyes, Jr., R., Ballenger, J.C. Lydiard, R.B., Sievers. G. and Mihalco, D. (1994) Predictors of response to alprazolam and placebo in patients with panic disorder. .I. Affect. Disord. 30, 5-13.