International Clinical Psychopharmacology:
Newer antidepressants and panic disorder: a meta-analysis
Andrisano, Costanzaa; Chiesa, Albertoa,b; Serretti, Alessandroa
aInstitute of Psychiatry, University of Bologna, Bologna, Italy
bSection of Pharmacology, Department of Clinical and Experimental Medicine and Pharmacology, University of Messina, Messina, Italy
All supplementary digital content is available directly from the corresponding author.
Correspondence to Alberto Chiesa, MD, Institute of Psychiatry, University of Bologna, Viale Carlo Pepoli 5, 40123 Bologna, Italy Tel: +39 051 6584233; fax: +39 051 521030; e-mail: firstname.lastname@example.org
Received April 26, 2012
Accepted September 12, 2012
Selective serotonin reuptake inhibitors and venlafaxine are currently considered as first-line agents for patients with panic disorder (PD). However, a systematic comparison of newer antidepressants for the treatment of PD is lacking thus far. Eligible studies focusing on PD patients treated with newer antidepressants were entered in the Cochrane Collaboration Review Manager. Our primary outcome measure was the mean change in panic symptoms from the baseline to the endpoint in patients treated with antidepressants as compared with those treated with placebo. Secondary outcome measures included the mean change in the overall anxiety scores and dropout rates. Sensitivity analyses were also carried out. Fifty studies focusing on 5236 patients were included. The following antidepressants were significantly superior to placebo for PD patients with the following increasing order of effectiveness: citalopram, sertraline, paroxetine, fluoxetine, and venlafaxine for panic symptoms and paroxetine, fluoxetine, fluvoxamine, citalopram, venlafaxine, and mirtazapine for overall anxiety symptoms. Aside from reboxetine and fluvoxamine, all drugs were associated with significantly lower dropout rates as compared with placebo. Several clinical variables moderated clinical outcomes. However, because of some inconsistencies across the studies and limited evidence for some drugs under investigation, further head-to-head comparisons are required.
Pharmacological options approved by current guidelines for the treatment of panic disorder (PD) include tricyclic antidepressant, benzodiazepines, monoamine oxidase inhibitors, selective serotonin reuptake inhibitors (SSRIs), and serotonin–norepinephrine reuptake inhibitors (Pollack et al., 2003a; Royal Australian and New Zealand College of Psychiatrists Clinical Practice Guidelines Team for Panic Disorder and Agoraphobia, 2003; Baldwin et al., 2005; Bandelow et al., 2008; American Psychiatric Association, 2009). With little differences among guidelines, SSRIs and the serotonin–norepinephrine reuptake inhibitor venlafaxine are currently considered as first-line agents for PD patients because of their favorable balance of efficacy and side effects’ profile. Tricyclic antidepressants and monoamine oxidase inhibitors should be considered as alternative choices only when patients do not seem to respond to or tolerate first-line treatments (Bandelow et al., 1995; Bandelow and Ruther, 2004; Bakker et al., 2005).
A critical evaluation of current guidelines, however, points to the dearth of information as to which pharmacological compound can be best tailored to the specific needs of each individual patient. Furthermore, although several reviews have been published recently that deal with the pharmacological approaches of PD (e.g. Mochcovitch and Nardi, 2010; Zobel and Maier, 2010; Batelaan et al., 2011; Freire et al., 2011), a systematic comparison of newer antidepressants (Cipriani et al., 2009) for the treatment of PD on the basis of a quantitative meta-analytic approach is lacking thus far. Therefore, the extent to which different antidepressants vary in terms of efficacy and acceptability is unclear. However, such information could provide clinicians with a useful tool for guiding clinical choices on the basis of rigorous empirical evidence and for possibly enhancing treatment outcomes.
Accordingly, the aim of this meta-analysis is to compare the short-term efficacy and tolerability of newer antidepressants for the treatment of PD. In addition, we aim to investigate whether the main outcomes of interest are moderated by a number of easily available clinical variables such as the duration of illness, the age of the patient, and the comorbidity with other psychiatric disorders.
A literature research was carried out using MEDLINE, ISI Web of Science, the Cochrane database, and references of retrieved articles. The research included articles written in English published up to January 2012. In line with a recent meta-analysis focusing on the efficacy and tolerability of newer antidepressants for patients with major depression (Cipriani et al., 2009), we focused on the same drugs for patients with PD. The main keywords were bupropion, citalopram, duloxetine, escitalopram, fluoxetine, fluvoxamine, milnacipran, mirtazapine, paroxetine, reboxetine, sertraline, and venlafaxine in combination with ‘panic disorder’.
Selection of trials
Two reviewers independently searched eligible articles for inclusion (A.C. and C.A.). The studies included had to: (a) focus on patients with PD (age>18) with or without further psychiatric comorbidities; (b) use one or more antidepressants mentioned above; (c) provide information about the efficacy of the antidepressant(s) under investigation on anxiety symptoms and/or on panic symptoms; and (d) have a minimum duration of 6 weeks. The following articles were excluded: (a) case reports and case series; (b) review articles and meta-analyses; (c) long-term studies (lasting >12 weeks) that did not provide information about the improvement in anxiety levels or panic symptoms in the short term (6–12 weeks); (d) studies investigating samples of patients treated with different antidepressants but that did not provide separate information about each specific antidepressant; (e) studies focusing on PD patients in remission at study entry; and (f) studies focusing on the combination of a pharmacological treatment with a nonpharmacological therapy of any kind. Studies with different methodological designs were included. However, sensitivity analyses were carried out to estimate the impact of study design on the outcomes of interest (see below).
Our primary outcome measure was the mean change in panic symptoms from the baseline to the endpoint in patients treated with antidepressants as compared with those treated with placebo. Secondary outcome measures included the mean change in the overall anxiety scores and dropout rates in patients treated with antidepressants as compared with those treated with placebo.
Scales specifically designed to measure panic symptoms (e.g. the Panic Disorder Severity Scale; Shear et al., 1997) were entered into the analysis specifically dealing with the severity of PD symptoms (primary outcome measure). Anxiety scales that were not specific for the assessment of PD symptoms [e.g. the Hamilton Rating Scale for Anxiety (Hamilton, 1959)] were included in the analysis focusing on overall anxiety levels.
Whenever it was possible, we referred to the 8-week treatment outcomes. If 8-week data were not available, we used data ranging between 6 and 12 weeks (we gave preference to the time-point considered in the original study as the endpoint). Patients who terminated the study before the endpoint for any reason were considered as dropouts.
Of note, in case of overlapping samples, the decision to include a given study was made on the basis of the following algorithm: (i) when no difference existed between two studies in terms of outcome measures, we considered the study with the largest sample size and (ii) when the studies provided different outcomes, we chose the study whose outcomes were more in line with the criteria of the present meta-analysis.
Data extraction and analysis
Data were independently extracted from original reports by two reviewers (A.C. and C.A.). The quality of the included article was assessed independently by the two reviewers using validated quality scales [the Jadad Scale (Jadad et al., 1996) for controlled studies and the Newcastle–Ottawa Scale (Wells et al., 2005) modified for uncontrolled studies for the remaining studies]. For controlled trials, a score of more than or equal to 3 was considered to be indicative of a high quality (Jadad et al., 1996), whereas for uncontrolled ones, a score of more than or equal to 4 was considered as indicative of high quality (Wells et al., 2005). All disagreements were resolved through discussion and, in case of persistent disagreement, with the involvement of a third reviewer (A.S.). No final disagreement was observed. All data were referred to the intention-to-treat populations, when available. Otherwise, the completers’ sample was analyzed. In addition, when one study included two (or more) treatment arms of patients treated with the same antidepressant at different dosages, we separately considered each treatment arm as if they were two (or more) independent studies.
Effect sizes of improvement were calculated by Hedges’s g (Hedges and Olkin, 1985). Because many studies included more than one assessment scale for the symptoms’ categories under investigation (panic symptoms and anxiety), we first calculated one effect size for each scale. Then, we calculated one weightened effect size for each type of measure (panic symptoms and anxiety symptoms) for each study (Mullen, 1989; Cohen, 1992; Shieh, 2012).
In addition, because better estimates can be derived when there are many samples and reasonable sample sizes (Hunter and Schmidt, 1990), in accordance with previous meta-analyses (e.g. Reger et al., 2004; Lopez-Leon et al., 2008), we focused only on drugs investigated in at least three independent studies. Indeed, although there are no clear guidelines on this topic, there are no statistical procedures to detect publication bias when the number of primary studies is low. Data for drugs investigated in at least three independent studies were entered into the Cochrane Collaboration Review Management Software (RevMan version 5.1; Cochrane Collaboration, Oxford, UK) and analyzed by RevMan analysis 1.01. For dichotomous outcomes, odds ratios and their 95% confidence intervals (CIs) were calculated, whereas for continuous outcomes, the standardized mean differences and their 95% CIs were calculated. For both dichotomous and continuous outcomes, a random-effect model was used, which takes into account possible differences in the implementation of intervention and the characteristics of the participants included.
Heterogeneity across the studies was assessed by the χ2 and Ι2 statistics and by visual inspection of the results. A χ2 statistic P less than 0.05 as well as an Ι2 statistic higher than 50% were considered to indicate heterogeneity. Moreover, for studies that did not include a placebo control group, we calculated the weighted mean of the placebo samples for the considered variable from placebo-controlled studies that investigated the same variable and applied it to studies not including a placebo group (Durrleman and Chaikin, 2003; Serretti and Mandelli, 2010). In such cases, the sample size of the virtual placebo group was considered as equal to the number of patients treated with antidepressants in the specific studies that did not include a placebo group to avoid inflation of the estimate. Also, for studies that did not provide the SD for continuous variables, we calculated the weighted mean of SDs provided by studies investigating the same measure and providing SD and we applied it to studies not providing such measures (Furukawa et al., 2006).
Additional sensitivity analyses were carried out on randomized-controlled studies, placebo-controlled studies, antidepressant monotherapy studies, nonsponsored studies (studies were considered as sponsored if they were funded, at least in part, by a pharmaceutical company or if at least one of the authors was a member of a pharmaceutical company), and studies focusing on PD patients without comorbidity to determine whether efficacy measures and dropout rates varied as a function of these variables. Moreover, the influence of several key variables on the main outcomes of interest, including (a) the mean dosage of antidepressants, (b) time considered as the endpoint in the present meta-analysis, (c) mean age of the sample, (d) percentage of female patients, (e) duration of illness, and (f) study’s quality, was calculated using the meta-regression model within the statistics environment R, version 2.8.1 (http://www.R-project.org), package ‘metaphor’. The existence of a possible publication bias was assessed by visual inspection of the funnel plots and by calculating quantitatively the influence of the publication bias by Egger’s analysis (Egger et al., 1997).
The original search identified 5169 papers. Five thousand and thirty-three studies were excluded because they did not examine the use of antidepressants in PD. After the inclusion and the exclusion criteria were applied to the remaining 136 studies, 86 studies were excluded and 50 studies could be included in the present meta-analysis (10 concerning citalopram, seven fluoxetine, eight fluvoxamine, five mirtazapine, 15 paroxetine, three reboxetine, four sertraline, and six venlafaxine; Fig. 1). Among the studies included, 26 studies used a randomized-controlled design, three studies used a nonrandomized-controlled design, and 21 studies used an uncontrolled design. A summary of the studies included is shown in Tables 1 and 2. Twenty-five out of 29 controlled studies and 10 out of 21 uncontrolled studies were considered to have a high-quality design (supplementary data 1). Bupropion, duloxetine, escitalopram, and milnacipran were excluded from the analyses because they were investigated in less than three studies.
Primary outcome measure
All antidepressants under investigation, with the exception of mirtazapine, reboxetine, and fluvoxamine, were significantly more efficacious than placebo on panic symptoms with the following increasing order of effectiveness: citalopram, sertraline, paroxetine, fluoxetine, and venlafaxine (Fig. 2 and supplementary data 2).
Secondary outcome measures
In terms of secondary outcomes, we found that, aside from reboxetine, and sertraline which were found not to significantly different from placebo, all drugs were effective in reducing anxiety levels (Fig. 2 and supplementary data 2). More in detail, we observed that paroxetine, fluoxetine, fluvoxamine, citalopram, venlafaxine and mirtazapine had an increasing effect on the reduction of anxiety symptoms, paroxetine being the drug with the smallest effect and mirtazapine the drug with the largest effect.
In terms of dropout rates, we found that they were significantly higher in patients treated with placebo as compared with patients treated with antidepressants, with the only exception of fluvoxamine and reboxetine (Fig. 2). More in detail, the following drugs were associated with decreasing dropout rates: venlafaxine, fluoxetine, sertraline, paroxetine, citalopram, and mirtazapine (supplementary data 2). Note, however, that a significant heterogeneity was observed in the analyses, indicating that clinical improvement and dropout rates showed large variations across the studies (supplementary data 2).
Improvement in panic symptoms
When we focused on randomized-controlled studies and on placebo-controlled studies separately, sertraline was no longer superior to placebo. In antidepressant monotherapy studies, we found that only fluoxetine and venlafaxine were significantly more effective than placebo. Moreover, paroxetine was the only drug superior to placebo in studies without a sponsor. Finally, with the only exception of reboxetine, all antidepressants under investigation were more effective than placebo in treating PD patients without comorbidities (supplementary data 3).
Improvement in anxiety levels
The use of different study designs had a significant impact on the efficacy of several antidepressants in the treatment of PD. In particular, when randomized-controlled studies were considered separately, we observed that mirtazapine, paroxetine, and sertraline were no longer superior to placebo. However, when we focused only on placebo-controlled studies, paroxetine was associated with a higher improvement in anxiety levels from the baseline to the endpoint compared with that observed in the general analysis, whereas the same was not true for sertraline (see supplementary data 3). When studies allowing only antidepressant monotherapy were considered separately, venlafaxine and paroxetine were no longer superior to placebo. Paroxetine was also not significantly different from placebo in nonsponsored studies. In addition, a sensitivity analysis focusing on studies including PD patients without comorbidities showed that mirtazapine, paroxetine, and venlafaxine were no more effective than placebo for the improvement in anxiety levels in this group of patients. For specific details, see supplementary data 3.
When data derived from randomized-controlled trials were analyzed separately, we observed that only venlafaxine and paroxetine were associated with significantly lower dropout rates as compared with placebo. Furthermore, the sensitivity analysis including only placebo-controlled studies showed that the number of dropouts did not differ significantly between patients treated with antidepressants and those treated with placebo, with the only exception of venlafaxine. Moreover, in studies allowing only antidepressant monotherapy, the dropout rates did not change significantly between patients treated with placebo and those treated with the following antidepressants: fluoxetine, fluvoxamine, and reboxetine. Also, when nonsponsored studies were considered separately, we found that all antidepressants under investigation were associated with significantly lower dropout rates as compared with placebo. In addition, the analysis focusing on PD patients without comorbidities showed that no significant difference was observed between fluoxetine, fluvoxamine, mirtazapine, and placebo in terms of dropout rates, whereas citalopram, paroxetine, reboxetine, venlafaxine, and sertraline were associated with a lower likelihood of dropout as compared with placebo (supplementary data 3).
Improvement in panic symptoms
We found that citalopram was more effective in alleviating panic symptoms if the duration of illness was shorter. Furthermore, paroxetine and sertraline showed a higher efficacy in studies with a longer duration. In addition, older age was significantly associated with higher clinical improvement in patients treated with sertraline. Finally, higher quality was positively correlated with PD symptoms’ improvement in patients treated with citalopram and negatively correlated with improvement in PD symptoms in patients taking venlafaxine.
Improvement in anxiety levels
We found that fluoxetine was more effective in alleviating anxiety levels when patients had a shorter duration of illness. Older age was significantly associated with higher clinical improvement in patients treated with fluvoxamine. Moreover, we observed that sertraline was more beneficial at lower doses. Finally, the efficacy of paroxetine and venlafaxine increased with increasing duration of treatment.
We observed that higher doses of sertraline were strongly associated with a greater number of dropouts. Moreover, we found that women were more likely to drop out when taking venlafaxine or paroxetine. In addition, patients treated with venlafaxine who had a longer duration of illness were more likely to discontinue the trial as compared with those with a shorter duration of illness. However, patients treated with paroxetine were more likely to drop out if they had a shorter duration of illness. Finally, we found that patients were more likely to drop out in shorter term studies using venlafaxine and in longer term studies using fluoxetine (see supplementary data 4).
The publication bias was investigated by visual inspection of the funnel plots (supplementary Figs 1–3) as well through Egger’s analysis (supplementary data 5). A significant publication bias was detected only in the analysis focusing on the usefulness of venlafaxine for the improvement in panic symptoms (β=0.98, P=0.0004, intercept=−15.01, 95% CI=−21.35 to −8.67). However, this publication bias is not entirely reliable because only a small number of studies provided information about the effects of venlafaxine on the improvement in panic symptoms.
The findings of our work lend support to the higher efficacy of most antidepressants under investigation over placebo for the reduction of panic symptoms and anxiety levels. Reboxetine was the only drug that was found to be ineffective for the treatment of both panic symptoms and anxiety levels and mirtazapine as well as fluvoxamine were found not to be useful in alleviating panic symptomatology. Overall, our findings are in agreement with those reported in previous reviews and meta-analysis (van Balkom et al., 1997; Otto et al., 2001; Bakker et al., 2002; Mitte, 2005). However, our meta-analysis is the first to provide information about a clear ranking of antidepressants’ efficacy for overall anxiety and PD symptoms investigated separately.
Of note, no clear advantage of any SSRIs over the others has been established definitely so far (Bakker et al., 2005). Therefore, our results strengthen current evidence by showing that paroxetine, fluoxetine, fluvoxamine, and citalopram could be associated with increasingly higher effectiveness in lowering anxiety levels. Of note, paroxetine was found to be the less effective antidepressant in the general analysis focusing on anxiety levels. However, this was not true in the analysis including only placebo-controlled studies. The discrepancy observed between these two analyses could be related to the large variety of methodological designs of the studies included focusing on paroxetine. Indeed, several of these studies used an uncontrolled design that has required the implementation of several virtual placebo samples. This could also explain why further sensitivity analyses focusing, as an example, on nonsponsored and monotherapy studies showed that paroxetine did not differ significantly from placebo. It should be noted, however, that in the sensitivity analysis of placebo-controlled studies specifically focusing on panic symptoms, paroxetine was associated with the highest improvement. This finding can be explained by the fact that paroxetine seems to have the highest affinity for the serotonin transporter as compared with other SSRIs (Owens and Nemeroff, 1994; Owens et al., 2008).
In terms of reboxetine, the only antidepressant that was found to be ineffective in treating both panic and anxiety symptoms in the general analysis, it is worth mentioning that only one study has been published so far that examined the efficacy of reboxetine in comparison with placebo for PD treatment (Versiani et al., 2002). The results of this study suggested that reboxetine was significantly more effective than placebo in reducing the final mean number of panic attacks and phobic symptoms. A significant limitation of this study was, however, the concomitant use of benzodiazepines that did not allow us to understand to what extent reboxetine alone was effective for PD patients. In addition, in the present meta-analysis, we found that reboxetine was ineffective for the treatment of PD patients not only in the general analysis but also in the sensitivity analysis focusing on antidepressant monotherapy. One possible explanation for this finding could be the fact that the serotonergic dysfunction is believed to play a more significant role in the etiology of PD as compared with the noradrenergic dysfunction (Maron and Shlik, 2006). Therefore, because reboxetine acts only on the noradrenergic system (Brunello and Racagni, 1998; Hajos et al., 2004), it might not be as effective for the treatment of PD patients as dual-acting antidepressants or as drugs inhibiting only the reuptake of serotonin. We underscore, however, that evidence deriving from high-quality studies on reboxetine is too little to draw definitive conclusions.
Our results also indicate that mirtazapine was no more effective than placebo for the reduction of panic symptoms. However, this newer antidepressant was associated with the highest improvement in anxiety levels. A possible explanation for such a discrepancy could be related to the fact that mirtazapine could be more effective for general anxiety symptoms such as worries, insomnia, and muscle tension, possibly because of its sedative profile, but less effective on specific panic symptoms. However, this explanation remains speculative thus far. Furthermore, no placebo-controlled trials of mirtazapine for the treatment of PD have been published thus far. Therefore, these results should be considered with extreme caution pending further replications in properly powered double-blind placebo-controlled studies.
Moreover, when studies allowing only antidepressants monotherapy were considered separately, only venlafaxine and fluoxetine were associated with a significant improvement in panic symptoms compared with placebo, indicating that patients treated with the other antidepressants could achieve higher benefits when such antidepressants are associated with anxiolytic drugs. Our results also suggested that paroxetine, mirtazapine, and venlafaxine did not significantly differ from placebo in terms of clinical improvement in anxiety when PD patients with comorbid psychiatric disorders were excluded. A possible explanation for these findings could be that clinical improvement in PD patients treated with these drugs could be largely because of the effects of these antidepressants on comorbid psychiatric disorders such as major depression that frequently occurs in comorbidity with PD. However, further in-depth studies are required to investigate this hypothesis.
In terms of the dropout rates, we found that a large number of antidepressants were associated with significantly lower dropout rates than placebo, fluvoxamine and reboxetine being notable exceptions. Furthermore, our data suggest that patients treated with mirtazapine were less likely to drop out as compared with patients treated with other drugs. It should be noted, however, that our data indicating that no significant difference was observed between reboxetine and placebo in terms of dropout rates are in conflict with some studies focusing on reboxetine (Dannon et al., 2002; Versiani et al., 2002; Seedat et al., 2003) showing that less than 5% of patients treated with reboxetine discontinued the trial. Moreover, our results differ from those reported by Wagner et al. (1994, 1996), who point to the good safety and tolerability profile of fluvoxamine. Similar to efficacy measures, however, it is worth mentioning that the discrepancy observed between our findings and those reported in earlier studies might be attributed to the large variety of methodological designs of the studies included.
Of note, the sensitivity analysis focusing on a placebo-controlled study found that there were no significant differences in terms of the dropout rates between patients treated with placebo and those treated with most antidepressants. A possible explanation for this finding could be that patients taking placebo may be more likely to drop out because of lack of efficacy, whereas patients taking active drugs could be more likely to drop out because of the occurrence of adverse effects. This explanation is consistent with observations reported in a large number of placebo-controlled studies focusing on the efficacy of antidepressants under investigation for patients with PD (e.g. Nair et al., 1996; Michelson et al., 1998).
In addition to the findings mentioned above, our meta-analysis suggested that several variables moderated the clinical outcomes and dropout rates. For example, we observed that sertraline was more beneficial on anxiety levels at lower doses, possibly because lower doses are better tolerated and less associated with activation symptoms. Indeed, we also found that higher doses of sertraline were strongly associated with a greater number of dropouts. Moreover, as reviewed by Pollack et al. (2000), the rate of dropouts because of adverse effects was higher in the fixed-dose studies (Gorman and Wolkow, 1994; Londborg et al., 1998), in which the initial dose level was higher and the titration schedule did not allow dose adjustments in response to emergent side effects, as compared with the flexible-dose studies (Pohl et al., 1998; Pollack et al., 1998).
In sum, our results preliminary suggest a possible specificity profile for drugs under investigation that require further investigation in properly powered placebo randomized-controlled trials. In case of replication, our results could have important clinical implications because they could allow us to match each patient to his/her most appropriate drug during the short-term treatment of PD and possibly enhance treatment outcomes.
Although the reviewed findings are noteworthy and could have important clinical implications, we emphasize that they should be considered with caution because of the following limitations. First, we included both controlled and uncontrolled studies. Furthermore, we included both double-blind, single-blind, and open trials, even though double-blind randomized-controlled studies showed superiority to other study designs and are usually preferred for meta-analytic purposes (Colditz et al., 1989; Schulz et al., 1995). Note, however, that several analyses were carried out and that, with the only exception of citalopram and venlafaxine, our meta-regression analysis focusing on the impact of study quality on clinical and safety outcomes did not provide evidence to suggest that it significantly affected the results. Moreover, it would be methodologically more rigorous to select only head-to-head trials, but we preferred to also include studies using different methodological designs to collect a larger number of studies. This could explain why a high heterogeneity was observed across the studies included in the present meta-analysis.
Second, for studies that did not include a placebo control group, we calculated a virtual placebo. The virtual control groups that we used to contrast drugs’ effectiveness could have been different from the population studied. Nevertheless, it is worth mentioning that in a previous meta-analysis, an analysis aimed at comparing the results observed in placebo-controlled trials separately analyzed with those derived including virtual placebo samples, no significant differences were found (Serretti and Mandelli, 2010). Furthermore, our strategy has the advantage to allow for the inclusion of more data that would otherwise be excluded. Third, we did not use a uniform measure to calculate the clinical improvement in panic and anxiety levels from the baseline to the endpoint. However, efficacy measures were analyzed by Hedge’s g. Hedge’s g standardizes the results of studies employing different assessment measures to a uniform scale so that they can be combined.
Fourth, we calculated missing SDs by calculating the weighted mean of SDs provided by studies investigating the same measure and we applied them to studies not providing this measure. Note, however, that this method has been validated empirically and seems adequate and reliable (Furukawa et al., 2006). Fifth, we included studies with a small sample size, which limits their power to detect differences among different treatment options. In any case, this critical issue is partially compensated by the weights of the sample size that reduce the weight of small trials. Then, an artificial inflation of the effect could be because of the use of the same normative control sample scores. Also, we did not include bupropion, duloxetine, escitalopram, and milnacipran because they were investigated in very few studies (Hunter and Schmidt, 1990). Finally, our analyses focused only on the acute-phase treatment of PD and cannot be generalized to very short-term periods (<6 weeks) or to periods longer than 12 weeks. Selecting the studies in the short-term treatment allowed us to include a greater number of pharmacological trials in our work. Moreover, it is worth mentioning that the majority of studies in the literature investigating the acute PD treatment report the eighth week as the endpoint.
The results of the present meta-analysis indicate that the majority of antidepressants under investigation are effective and safe for the short-term treatment of PD. Also, our meta-analysis is the first to provide information on a clear ranking of antidepressants’ efficacy for overall anxiety and PD symptoms investigated separately and to provide information on the possible moderators of outcome. However, considering the several limitations stated above, our results should be considered with caution and point to the strong need for more rigorous and properly powered double-blind head-to-head comparison trials that could provide more reliable data on how different drugs differ in terms of efficacy and tolerability for the treatment of PD.
Conflicts of interest
There are no conflicts of interest.
Amore M, Magnani K, Cerisoli M, Casagrande C, Ferrari G. Panic disorder. A long-term treatment study: fluoxetine vs imipramine. Hum Psychopharmacol. 1999;14:429–434
Bakker A, van Balkom AJ, Spinhoven P. SSRIs vs. TCAs in the treatment of panic disorder: a meta-analysis. Acta Psychiatr Scand. 2002;106:163–167
Bakker A, van Balkom AJ, Stein DJ. Evidence-based pharmacotherapy of panic disorder. Int J Neuropsychopharmacol. 2005;8:473–482
Baldwin DS, Anderson IM, Nutt DJ, Bandelow B, Bond A, Davidson JR, et al. Evidence-based guidelines for the pharmacological treatment of anxiety disorders: recommendations from the British Association for Psychopharmacology. J Psychopharmacol. 2005;19:567–596
Ballenger JC, Wheadon DE, Steiner M, Bushnell W, Gergel IP. Double-blind, fixed-dose, placebo-controlled study of paroxetine in the treatment of panic disorder. Am J Psychiatry. 1998;155:36–42
Bandelow B, Ruther E. Treatment-resistant panic disorder. CNS Spectr. 2004;9:725–739
Bandelow B, Sievert K, Rothemeyer M, Hajak G, Ruther E. What treatments do patients with panic disorder and agoraphobia get? Eur Arch Psychiatry Clin Neurosci. 1995;245:165–171
Bandelow B, Behnke K, Lenoir S, Hendriks GJ, Alkin T, Goebel C, et al. Sertraline versus paroxetine in the treatment of panic disorder: an acute, double-blind noninferiority comparison. J Clin Psychiatry. 2004;65:405–413
Bandelow B, Zohar J, Hollander E, Kasper S, Moller HJ, Zohar J, et al. World Federation of Societies of Biological Psychiatry (WFSBP) guidelines for the pharmacological treatment of anxiety, obsessive-compulsive and post-traumatic stress disorders – first revision. World J Biol Psychiatry. 2008;9:248–312
Batelaan NM, Van Balkom AJ, Stein DJ. Evidence-based pharmacotherapy of panic disorder: an update. Int J Neuropsychopharmacol. 2011:1–13
Bertani A, Perna G, Migliarese G, Di Pasquale D, Cucchi M, Caldirola D, et al. Comparison of the treatment with paroxetine and reboxetine in panic disorder: a randomized, single-blind study. Pharmacopsychiatry. 2004;37:206–210
Black DW, Wesner R, Bowers W, Gabel J. A comparison of fluvoxamine, cognitive therapy, and placebo in the treatment of panic disorder. Arch Gen Psychiatry. 1993;50:44–50
Boshuisen ML, Slaap BR, Vester-Blokland ED, den Boer JA. The effect of mirtazapine in panic disorder: an open label pilot study with a single-blind placebo run-in period. Int Clin Psychopharmacol. 2001;16:363–368
Brunello N, Racagni G. Rationale for the development of noradrenaline reuptake inhibitors. Hum Psychopharmacol. 1998;13:13–19
Carli V, Sarchiapone M, Camardese G, Romano L, DeRisio S. Mirtazapine in the treatment of panic disorder. Arch Gen Psychiatry. 2002;59:661–662
Cipriani A, Furukawa TA, Salanti G, Geddes JR, Higgins JP, Churchill R, et al. Comparative efficacy and acceptability of 12 new-generation antidepressants: a multiple-treatments meta-analysis. Lancet. 2009;373:746–758
Cohen J. A power primer. Psychol Bull. 1992;112:155–159
Colditz GA, Miller JN, Mosteller F. How study design affects outcomes in comparisons of therapy. I: medical. Stat Med. 1989;8:441–454
Coplan JD, Papp LA, Martinez J, Pine D, Rosenblum LA, Cooper T, et al. Persistence of blunted human growth hormone response to clonidine in fluoxetine-treated patients with panic disorder. Am J Psychiatry. 1995;152:619–622
Dannon PN, Iancu I, Grunhaus L. The efficacy of reboxetine in the treatment-refractory patients with panic disorder: an open label study. Hum Psychopharmacol. 2002;17:329–333
Den Boer JA, Westenberg HG. Effect of a serotonin and noradrenaline uptake inhibitor in panic disorder; a double-blind comparative study with fluvoxamine and maprotiline. Int Clin Psychopharmacol. 1988;3:59–74
Den Boer JA, Westenberg HG. Serotonin function in panic disorder: a double blind placebo controlled study with fluvoxamine and ritanserin. Psychopharmacology (Berl). 1990;102:85–94
Durrleman S, Chaikin P. The use of putative placebo in active control trials: two applications in a regulatory setting. Stat Med. 2003;22:941–952
Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ. 1997;315:629–634
Ferguson JM, Khan A, Mangano R, Entsuah R, Tzanis E. Relapse prevention of panic disorder in adult outpatient responders to treatment with venlafaxine extended release. J Clin Psychiatry. 2007;68:58–68
Freire RC, Hallak JE, Crippa JA, Nardi AE. New treatment options for panic disorder: clinical trials from 2000 to 2010. Expert Opin Pharmacother. 2011;12:1419–1428
Furukawa TA, Barbui C, Cipriani A, Brambilla P, Watanabe N. Imputing missing standard deviations in meta-analyses can provide accurate results. J Clin Epidemiol. 2006;59:7–10
Goddard AW, Brouette T, Almai A, Jetty P, Woods SW, Charney D. Early coadministration of clonazepam with sertraline for panic disorder. Arch Gen Psychiatry. 2001;58:681–686
Gorman J, Wolkow R. Sertraline as a treatment for panic disorder. Neuropsychopharmacology. 1994;10:117–175
Gulsun M, Doruk A, Uzun O, Turkbay T, Ozsahin A. Effect of dissociative experiences on drug treatment of panic disorder. Clin Drug Investig. 2007;27:583–590
Hajos M, Fleishaker JC, Filipiak-Reisner JK, Brown MT, Wong EH. The selective norepinephrine reuptake inhibitor antidepressant reboxetine: pharmacological and clinical profile. CNS Drug Rev. 2004;10:23–44
Hamilton M. The assessment of anxiety states by rating. Br J Med Psychol. 1959;32:50–55
Hedges LV, Olkin I Statistical methods for meta-analysis. 1985 Orlando, FL Academic Press
Hoehn-Saric R, McLeod DR, Hipsley PA. Effect of fluvoxamine on panic disorder. J Clin Psychopharmacol. 1993;13:321–326
Humble M, Wistedt B. Serotonin, panic disorder and agoraphobia: short-term and long-term efficacy of citalopram in panic disorders. Int Clin Psychopharmacol. 1992;6(Suppl 5):21–39
Hunter JE, Schmidt FL Methods of meta-analysis. Correcting error and bias in research finding. 1990 Beverly Hills, CA Sage Publications
Jadad AR, Moore RA, Carroll D. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Controlled Clin Trials. 1996;17:1–12
Kamijima K, Kuboki T, Kumano H, Burt T, Cohen G, Arano I, et al. A placebo-controlled, randomized withdrawal study of sertraline for panic disorder in Japan. Int Clin Psychopharmacol. 2005;20:265–273
Kim EJ, Yu BH. Increased cholesterol levels after paroxetine treatment in patients with panic disorder. J Clin Psychopharmacol. 2005;25:597–599
Kindler S, Dolberg OT, Cohen H, Hirschmann S, Kotler M. The treatment of comorbid premature ejaculation and panic disorder with fluoxetine. Clin Neuropharmacol. 1997;20:466–471
Lepola U, Leinonen E, Turtonen J, Pentinen J. The effect of citalopram in panic disorder and agoraphobia. Nord J Psychiatry. 1994:13–17
Liebowitz MR, Asnis G, Mangano R, Tzanis E. A double-blind, placebo-controlled, parallel-group, flexible-dose study of venlafaxine extended release capsules in adult outpatients with panic disorder. J Clin Psychiatry. 2009;70:550–561
Londborg PD, Wolkow R, Smith WT, DuBoff E, England D, Ferguson J, et al. Sertraline in the treatment of panic disorder. A multi-site, double-blind, placebo-controlled, fixed-dose investigation. Br J Psychiatry. 1998;173:54–60
Lopez-Leon S, Janssens AC, Gonzalez-Zuloeta Ladd AM, Del-Favero J, Claes SJ, Oostra BA, et al. Meta-analyses of genetic studies on major depressive disorder. Mol Psychiatry. 2008;13:772–785
Maron E, Shlik J. Serotonin function in panic disorder: important, but why? Neuropsychopharmacology. 2006;31:1–11
Michelson D, Lydiard RB, Pollack MH, Tamura RN, Hoog SL, Tepner R, et al. Outcome assessment and clinical improvement in panic disorder: evidence from a randomized controlled trial of fluoxetine and placebo. The Fluoxetine Panic Disorder Study Group. Am J Psychiatry. 1998;155:1570–1577
Michelson D, Allgulander C, Dantendorfer K, Knezevic A, Maierhofer D, Micev V, et al. Efficacy of usual antidepressant dosing regimens of fluoxetine in panic disorder: randomised, placebo-controlled trial. Br J Psychiatry. 2001;179:514–518
Mitte K. A meta-analysis of the efficacy of psycho- and pharmacotherapy in panic disorder with and without agoraphobia. J Affect Disord. 2005;88:27–45
Mochcovitch MD, Nardi AE. Selective serotonin-reuptake inhibitors in the treatment of panic disorder: a systematic review of placebo-controlled studies. Expert Rev Neurother. 2010;10:1285–1293
Montanes-Rada F, de Lucas-Taracena MT, Sanchez-Romero S. Mirtazapine versus paroxetine in panic disorder: an open study. Int J Psych Clin Pract. 2005;9:87–93
Mullen BHillsdale NE. Advanced BASIC meta-analysis. Advanced BASIC meta-analysis. 1989 Hillsdale, NJ Erlbaum
Nair NP, Bakish D, Saxena B, Amin M, Schwartz G, West TE. Comparison of fluvoxamine, imipramine, and placebo in the treatment of outpatients with panic disorder. Anxiety. 1996;2:192–198
Nardi AE, Valenca AM, Freire RC, Mochcovitch MD, Amrein R, Sardinha A, et al. Psychopharmacotherapy of panic disorder: 8-week randomized trial with clonazepam and paroxetine. Braz J Med Biol Res. 2011;44:366–373
Neuger J, Wistedt B, Sinner B, Aberg-Wistedt A, Stain-Malmgren R. The effect of citalopram treatment on platelet serotonin function in panic disorders. Int Clin Psychopharmacol. 2000;15:83–91
Oehrberg S, Christiansen PE, Behnke K, Borup AL, Severin B, Soegaard J, et al. Paroxetine in the treatment of panic disorder. A randomised, double-blind, placebo-controlled study. Br J Psychiatry. 1995;167:374–379
Otto MW, Tuby KS, Gould RA, McLean RY, Pollack MH. An effect-size analysis of the relative efficacy and tolerability of serotonin selective reuptake inhibitors for panic disorder. Am J Psychiatry. 2001;158:1989–1992
Owens MJ, Nemeroff CB. Role of serotonin in the pathophysiology of depression: focus on the serotonin transporter. Clin Chem. 1994;40:288–295
Owens MJ, Krulewicz S, Simon JS, Sheehan DV, Thase ME, Carpenter DJ, et al. Estimates of serotonin and norepinephrine transporter inhibition in depressed patients treated with paroxetine or venlafaxine. Neuropsychopharmacology. 2008;33:3201–3212
Papp LA, Sinha SS, Martinez JM, Coplan JD, Amchin J, Gorman JM. Low-dose venlafaxine treatment in panic disorder. Psychopharmacol Bull. 1998;34:207–209
Pecknold JC, Luthe L, Iny L, Ramdoyal D. Fluoxetine in panic disorder: pharmacologic and tritiated platelet imipramine and paroxetine binding study. J Psychiatry Neurosci. 1995;20:193–198
Pelland ME, Marchand A, Lessard MJ, Belleville G, Chauny JM, Vadeboncoeur A, et al. Efficacy of 2 interventions for panic disorder in patients presenting to the ED with chest pain. Am J Emerg Med. 2010;29:1051–1061
Perna G, Bertani A, Caldirola D, Smeraldi E, Bellodi L. A comparison of citalopram and paroxetine in the treatment of panic disorder: a randomized, single-blind study. Pharmacopsychiatry. 2001;34:85–90
Perna G, Alpini D, Caldirola D, Raponi G, Cesarani A, Bellodi L. Serotonergic modulation of the balance system in panic disorder: an open study. Depress Anxiety. 2003;17:101–106
Perna G, Favaron E, Di Bella D, Bussi R, Bellodi L. Antipanic efficacy of paroxetine and polymorphism within the promoter of the serotonin transporter gene. Neuropsychopharmacology. 2005;30:2230–2235
Pohl RB, Wolkow RM, Clary CM. Sertraline in the treatment of panic disorder: a double-blind multicenter trial. Am J Psychiatry. 1998;155:1189–1195
Pollack MH, Worthington JJ 3rd, Otto MW, Maki KM, Smoller JW, Manfro GG, et al. Venlafaxine for panic disorder: results from a double-blind, placebo-controlled study. Psychopharmacol Bull. 1996;32:667–670
Pollack MH, Otto MW, Worthington JJ, Manfro GG, Wolkow R. Sertraline in the treatment of panic disorder: a flexible-dose multicenter trial. Arch Gen Psychiatry. 1998;55:1010–1016
Pollack MH, Rapaport MH, Clary CM, Mardekian J, Wolkow R. Sertraline treatment of panic disorder: response in patients at risk for poor outcome. J Clin Psychiatry. 2000;61:922–927
Pollack MH, Allgulander C, Bandelow B, Cassano GB, Greist JH, Hollander E, et al. WCA recommendations for the long-term treatment of panic disorder. CNS Spectr. 2003a;8:17–30
Pollack MH, Simon NM, Worthington JJ, Doyle AL, Peters P, Toshkov F, et al. Combined paroxetine and clonazepam treatment strategies compared to paroxetine monotherapy for panic disorder. J Psychopharmacol. 2003b;17:276–282
Pollack M, Mangano R, Entsuah R, Tzanis E, Simon NM, Zhang Y. A randomized controlled trial of venlafaxine ER and paroxetine in the treatment of outpatients with panic disorder. Psychopharmacology (Berl). 2007a;194:233–242
Pollack MH, Lepola U, Koponen H, Simon NM, Worthington JJ, Emilien G, et al. A double-blind study of the efficacy of venlafaxine extended-release, paroxetine, and placebo in the treatment of panic disorder. Depress Anxiety. 2007b;24:1–14
Pols HJ, Hauzer RC, Meijer JA, Verburg K, Griez EJ. Fluvoxamine attenuates panic induced by 35% CO2 challenge. J Clin Psychiatry. 1996;57:539–542
Rampello L, Alvano A, Raffaele R, Malaguarnera M, Vecchio I. New possibilities of treatment for panic attacks in elderly patients: escitalopram versus citalopram. J Clin Psychopharmacol. 2006;26:67–70
Reger MA, Welsh RK, Watson GS, Cholerton B, Baker LD, Craft S. The relationship between neuropsychological functioning and driving ability in dementia: a meta-analysis. Neuropsychology. 2004;18:85–93
Ribeiro L, Busnello JV, Kauer-Sant’Anna M, Madruga M, Quevedo J, Busnello EA, et al. Mirtazapine versus fluoxetine in the treatment of panic disorder. Braz J Med Biol Res. 2001;34:1303–1307
. Australian and New Zealand clinical practice guidelines for the treatment of panic disorder and agoraphobia. Aust NZ J Psychiatry. 2003;37:641–656
Sarchiapone M, Amore M, De Risio S, Carli V, Faia V, Poterzio F, et al. Mirtazapine in the treatment of panic disorder: an open-label trial. Int Clin Psychopharmacol. 2003;18:35–38
Schulz KF, Chalmers I, Hayes RJ, Altman DG. Empirical evidence of bias. Dimensions of methodological quality associated with estimates of treatment effects in controlled trials. JAMA. 1995;273:408–412
Seedat S, van Rheede van Oudtshoorn E, Muller JE, Mohr N, Stein DJ. Reboxetine and citalopram in panic disorder: a single-blind, cross-over, flexible-dose pilot study. Int Clin Psychopharmacol. 2003;18:279–284
Serretti A, Mandelli L. Antidepressants and body weight: a comprehensive review and meta-analysis. J Clin Psychiatry. 2010;71:1259–1272
Sharp DM, Power KG, Simpson RJ. Fluvoxamine, placebo, and cognitive behavioural therapy used alone and in combination in the treatment of panic disorder and agoraphobia. J Anxiety Disord. 1996;10:219–242
Shear MK, Brown TA, Barlow DH, Money R, Sholomskas DE, Woods SW, et al. Multicenter collaborative panic disorder severity scale. Am J Psychiatry. 1997;154:1571–1575
Shieh G. Confidence intervals and sample size calculations for the weighted eta-squared effect sizes in one-way heteroscedastic ANOVA. Behav Res Methods. 2012
Shlik J, Aluoja A, Vasar V, Vasar E, Podar T, Bradwejn J. Effects of citalopram treatment on behavioural, cardiovascular and neuroendocrine response to cholecystokinin tetrapeptide challenge in patients with panic disorder. J Psychiatry Neurosci. 1997;22:332–340
Sim HB, Kang EH, Yu BH. Changes in cerebral cortex and limbic brain functions after short-term paroxetine treatment in panic disorder: an [F]FDG-PET Pilot Study. Psychiatry Invest. 2010;7:215–219
Stahl SM, Gergel I, Li D. Escitalopram in the treatment of panic disorder: a randomized, double-blind, placebo-controlled trial. J Clin Psychiatry. 2003;64:1322–1327
Van Balkom AJ, Bakker A, Spinhoven P, Blaauw BM, Smeenk S, Ruesink B. A meta-analysis of the treatment of panic disorder with or without agoraphobia: a comparison of psychopharmacological, cognitive-behavioral, and combination treatments. J Nerv Ment Dis. 1997;185:510–516
Van Vliet IM, den Boer JA, Westenberg HG, Slaap BR. A double-blind comparative study of brofaromine and fluvoxamine in outpatients with panic disorder. J Clin Psychopharmacol. 1996;16:299–306
Versiani M, Cassano G, Perugi G, Benedetti A, Mastalli L, Nardi A, et al. Reboxetine, a selective norepinephrine reuptake inhibitor, is an effective and well-tolerated treatment for panic disorder. J Clin Psychiatry. 2002;63:31–37
Wade AG, Lepola U, Koponen HJ, Pedersen V, Pedersen T. The effect of citalopram in panic disorder. Br J Psychiatry. 1997;170:549–553
Wagner W, Zaborny BA, Gray TE. Fluvoxamine. A review of its safety profile in world-wide studies. Int Clin Psychopharmacol. 1994;9:223–227
Wagner W, Hauser V, Wong LF. The safety profile of fluvoxamine in elderly patients. Hum Psychopharmacol. 1996;11:267–272
Zobel A, Maier W. Pharmacogenetics of antidepressive treatment. Eur Arch Psychiatry Clin Neurosci. 2010;260:407–417
antidepressants; mirtazapine; panic disorder; reboxetine; selective serotonin reuptake inhibitors; venlafaxine
© 2013 Lippincott Williams & Wilkins, Inc.
What does "Remember me" mean?
By checking this box, you'll stay logged in until you logout. You'll get easier access to your articles, collections,
media, and all your other content, even if you close your browser or shut down your
To protect your most sensitive data and activities (like changing your password),
we'll ask you to re-enter your password when you access these services.
What if I'm on a computer that I share with others?
If you're using a public computer or you share this computer with others, we recommend
that you uncheck the "Remember me" box.
Highlight selected keywords in the article text.
Data is temporarily unavailable. Please try again soon.