Advertisement
Research Article

Placebo Response of Non-Pharmacological and Pharmacological Trials in Major Depression: A Systematic Review and Meta-Analysis

  • André Russowsky Brunoni,

    Affiliations: Berenson-Allen Center for Noninvasive Brain Stimulation, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts, United States of America, Department and Institute of Psychiatry, University of Sao Paulo, Sao Paulo, Brazil

    X
  • Mariana Lopes,

    Affiliation: Berenson-Allen Center for Noninvasive Brain Stimulation, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts, United States of America

    X
  • Ted J. Kaptchuk,

    Affiliation: Osher Research Center, Harvard Medical School, Boston, Massachusetts, United States of America

    X
  • Felipe Fregni mail

    ffregni@bidmc.harvard.edu

    Affiliation: Berenson-Allen Center for Noninvasive Brain Stimulation, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts, United States of America

    X
  • Published: March 18, 2009
  • DOI: 10.1371/journal.pone.0004824

Abstract

Background

Although meta-analyses have shown that placebo responses are large in Major Depressive Disorder (MDD) trials; the placebo response of devices such as repetitive transcranial magnetic stimulation (rTMS) has not been systematically assessed. We proposed to assess placebo responses in two categories of MDD trials: pharmacological (antidepressant drugs) and non-pharmacological (device- rTMS) trials.

Methodology/Principal Findings

We performed a systematic review and meta-analysis of the literature from April 2002 to April 2008, searching MEDLINE, Cochrane, Scielo and CRISP electronic databases and reference lists from retrieved studies and conference abstracts. We used the keywords placebo and depression and escitalopram for pharmacological studies; and transcranial magnetic stimulation and depression and sham for non-pharmacological studies. All randomized, double-blinded, placebo-controlled, parallel articles on major depressive disorder were included. Forty-one studies met our inclusion criteria - 29 in the rTMS arm and 12 in the escitalopram arm. We extracted the mean and standard values of depression scores in the placebo group of each study. Then, we calculated the pooled effect size for escitalopram and rTMS arm separately, using Cohen's d as the measure of effect size. We found that placebo response are large for both escitalopram (Cohen's d - random-effects model - 1.48; 95%C.I. 1.26 to 1.6) and rTMS studies (0.82; 95%C.I. 0.63 to 1). Exploratory analyses show that sham response is associated with refractoriness and with the use of rTMS as an add-on therapy, but not with age, gender and sham method utilized.

Conclusions/Significance

We confirmed that placebo response in MDD is large regardless of the intervention and is associated with depression refractoriness and treatment combination (add-on rTMS studies). The magnitude of the placebo response seems to be related with study population and study design rather than the intervention itself.

Introduction

Placebo effect plays a significant role in clinical trials of major depressive disorder (MDD); in fact, two recent meta-analyses showed that the mean responder raters in the placebo group in antidepressant trials are 29.7% [1] and that drug-placebo differences might be relatively small in patients with MDD due to the large placebo response [2]. Given the importance of placebo response in MDD trials and the need to develop efficient research designs, it is critical to enhance our understanding on the placebo effects of distinct treatments such as repetitive transcranial magnetic stimulation (rTMS), a novel non-pharmacological intervention for neuropsychiatric diseases.

Several meta-analyses of rTMS clinical trials have been performed in the past ten years, initially showing negative or poor results [3], [4]; although two recent studies have demonstrated a greater efficacy of the method [5], [6]. However, taking into account the heterogeneity of rTMS trials and the lack of precise predictors of outcome, Herrmann and Ebmeier [7] proposed that non-specific contextual effects - such as the use of a new and relatively unknown technological device and the running of trials in major universities and teaching hospitals - play an important role in rTMS depression improvement. In fact, non-pharmacological treatments might have a large placebo response [8]. Furthermore, despite several meta-analyses assessing the placebo response of pharmacological trials in depression [1], [2], [9], [10] placebo response of transcranial magnetic stimulation has not been sufficiently explored. We therefore decided to assess the placebo response of such intervention and perform an exploratory comparison with a non-pharmacological intervention trough a systematic review and meta-analysis of recent clinical trials of major depression.

Aims of the study

This study sought to ascertain the magnitude of placebo response in controlled trials of rTMS and non-pharmacological studies using escitalopram as the antidepressant drug. Our secondary aim, given the limitations for such aim, was to exploratory compare the effect sizes of placebo responses of rTMS studies and pharmacological studies. The importance of our study is contribute towards a better understanding of the placebo effects mechanisms by comparing a traditional pill-taking medical ritual to a new sham-device healing context.

Methods

We chose escitalopram to estimate the placebo response of pharmacological treatment as several placebo-controlled trials have been recently conducted and for non-pharmacological treatment we chose rTMS as, similarly, several sham-controlled studies have also been performed recently. We performed a systematic review on all escitalopram and rTMS trials published since 2002 and subsequently performed two main analyses: for the placebo-drug response and for the sham-rTMS response. We then compared the effect size of these groups. We also performed exploratory analyses to assess predictors associated with placebo response.

We choose this time period because the first escitalopram trial was published in 2002 and we looked for concurrent rTMS and escitalopram trials to make the studies more comparable methodologically (i.e., with comparable sample sizes, diagnostic definitions, rating methods and quality of studies) and also because a meta-analysis performed in 2003 [3] stated that rTMS trials up to 2002 had been of low quality.

Literature Search

We searched for published articles from April 2002 to April 2008 (period of 96 months) in the following databases: MEDLINE, Web of Science, Cochrane, and SCIELO. We also examined reference lists in systematic reviews and retrieved papers. To check for unpublished trials, we: (i) consulted the CRISP database and the websites clinicaltrials.gov and clinicalstudyresults.org; (ii) contacted experts; (iii) searched for conference poster abstracts; (iv) searched for studies in the monograph reference lists of Lexapro ® and; (iv) sent e-mails asking for unpublished studies to Forest Labs and to Lundbeck S/A. Our key search terms were “depression”, “escitalopram”, and “placebo” in the escitalopram arm; and “depression”, “transcranial magnetic stimulation” and “sham” in the rTMS arm.

Selection criteria

The following inclusion criteria were adopted: (i) manuscript written in English (although there were no manuscripts in other languages); (ii) randomized, double-blinded, placebo-controlled (or sham-controlled), parallel studies on major depressive disorder; (iii) mood effects assessed by a continuous mood scale, such as Hamilton Depression Rating Scale (HDRS) or Montgomery-Asberg Depression Rating Scale (MADRS); (iv) studies that reported mean and standard deviation of the mood scales (or provided other statistical parameters that could be used to deduce this values) for the placebo group and; (v) studies published from April 2002 to April 2008.

Data extraction

Data were extracted independently by the first author (AB) and double-checked by the second author (ML), using a structured form. The discrepancies were resolved by consensus and the corresponding author (FF) consulted if needed. The following variables were extracted: 1) mean and standard deviation values of depression rating scales at baseline and end of treatment in active (active group was used for exploratory analysis) and placebo/sham groups and; 2) demographic, clinical and treatment characteristics (e.g. number of patients, age, gender, previous use of medications, depression-resistant subjects, duration of treatment, sham procedure utilized).

When the study did not report mean and standard deviation (SD) values, we either deduced them (using statistical parameters) or contacted the corresponding author. Many escitalopram studies did not report SD final scores – in these cases, we calculated SD from standard error (SE) at end-of-treatment or from SD or SE difference changes when possible [11], [12]. In two studies, SD had been only reported in graphs and we asked for data from Forest Research Institute [13], [14]. We also received data from Forest Labs of two posters [15], [16] and for an unpublished trial mentioned in another study [13]. Two authors failed to provide the requested data [17], [18]; in these cases we had to input SD post-treatment scores based on the mean of the available SD scores of other trials, a method suggested by The Cochrane Collaboration to be applied in such cases [19]. In the rTMS arm, some trials just reported data in graphs, while others did not report SD post-treatment scores. We contacted the corresponding authors in these cases [20], [21], [22], [23]. Many rTMS trials also reported several depression scores at different times using more than one depression scale -in such cases, we extracted the data presented by the authors as the main result. Finally, we used only unadjusted rating scores in our analysis.

Quality assessment

We looked for the following biases: (1) selection bias - adequate concealment of treatment (e.g., randomization was performed by lottery and sealed, opaque envelopes were used); (2) performance bias – if the study is single-blinded or double-blinded - for rTMS studies we checked if they were single-blinded studies with external blind raters and also if blinding of patients and physicians were assessed; (3) attrition bias – if data are adequately reported in the study, if there is evidence of intention-to-treat treatment, and if methods used to handle with missing data (e.g., last observation carried forward, complete case analysis) were reported.

Quantitative analysis

All of our analyses were performed using STATA statistical software, version 9.0 (Statacorp, College Station, TX, USA). We initially calculated the standardized mean difference and the pooled standard deviation for each comparison –i.e. for each study we calculated the change of either placebo or sham scores (baseline minus post-treatment scores) and divided by the standard deviation of change. We used Cohen's d as a measure of the effect size. Then, we measured the pooled weighted effect size (weighted by the inverse variance of each study) using the random and fixed effect models. We performed the analyses of placebo response in escitalopram and rTMS trials separately and further compared the pooled effect sizes. Heterogeneity was evaluated with Chi-square test. We also performed sensitivity analysis, cumulative regression and assessed publication bias using Begg-modified funnel plot and Egger test [24] for each analysis.

Meta-regression was performed using the random-effects model and tau2 variance was calculated by the method of the residual maximum likelihood. We tested the following variables: age (years), gender (%females), duration of treatment (weeks), and depression response in the active groups (Cohen's d pooled effect size of the active groups) – treated as continuous variables; sham procedure, treatment resistant patients (defined as more than 50% of patients failed at last two antidepressant treatments); drug-free patients; and rTMS as an add-on therapy were treated as categorical variables. It should be underscored that we classified as “angled coil” studies that described the use of an active rTMS coil in a different angle or position when applied to the scalp; whereas “sham coil” included studies that used a non-active coil associated with a method to preserve blindness (e.g. a study [25] described that sham stimulation was performed with “an identical coil (…) but without any electronic connection. This set-up had a similar sound effect but with no stimulation…”). Three studies used a different sham approach and were not pooled together in this analysis, because either a shielded coil [26] or a special coil generating a small field [23] were used.

Also, we considered as “add-on therapy” when a drug treatment was initiated simultaneously to active or sham rTMS, i.e., patients from sham group were actually starting an active drug treatment -in fact, this is the same concept of an “accelerating” study [27].

For baseline depression, we meta-regressed using either MADRS or HDRS baseline scores in escitalopram and rTMS trials, respectively. For rTMS studies that used MADRS scores as the primary outcome, we used the values of HDRS scores reported in secondary outcomes when this was possible [26], [28], [29], [30], [31]; in four studies this was not possible [20], [32], [33], [34] and therefore we imputed missing HDRS scores regressing for other variables. Finally we assessed whether improvement in the active group was correlated with the placebo response – including this variable in our model.

Results

Using the keywords previously mentioned we were able to find 67 citations for escitalopram and 92 for rTMS studies. Only 12 and 29 studies met our inclusion criteria. Reasons for exclusion included: (1) reviews and meta-analyses; (2) studies that assessed other psychiatric diseases; (3) other studies designs (open-label, cross-over designs, quasi-randomized trials); (4) lack of sham or placebo group; (5) other topics. (Fig. 1)

thumbnail

Figure 1. QUOROM flow chart used to identify studies for detailed analysis.

doi:10.1371/journal.pone.0004824.g001

Regarding study quality, all escitalopram studies are multi-centric, randomized (although only one study reports the allocation method), double-blinded, and performed an intention-to-treat analysis (ITT), using the last observation carrier forward (LOCF) method. The quality of the rTMS studies is heterogeneous: all studies are randomized (thirteen studies report the allocation method); and single-blinded with external evaluation but only 8 studies addressed the integrity of blinding. Eighteen studies performed an intention-to-treat analysis, while 11 performed a complete-case analysis – mostly, exploratory studies. Finally, only two rTMS studies are multicentric. The quality assessment of each study is reported in Table S1.

The clinical characteristics of the 41 studies are summarized in Table 1. Tables 2 and 3 show characteristics of each study. There were 680 patients in sham group in the 29 rTMS studies (median per study = 16, interquartile range (IQR) = 10–26), while the 12 escitalopram studies enrolled 1714 patients in the placebo group (median per study = 133, IQR = 128–153). Also, all escitalopram studies enrolled non-treatment resistant patients who were drug-free, while most patients in rTMS studies were refractory and using antidepressant drugs – in fact, in 6 studies an antidepressant drug was initiated in both active and sham groups at the beginning of the trial. Conversely, the groups were comparable regarding age (50.7 vs. 43.1 years), gender (59% vs. 61% females) and baseline HDRS (24.73 vs. 21.4) and MADRS (33.1 vs. 29.23) scores.

thumbnail

Table 1. General characteristics of the studies.

doi:10.1371/journal.pone.0004824.t001
thumbnail

Table 2. Characteristics of each rTMS study included.

doi:10.1371/journal.pone.0004824.t002
thumbnail

Table 3. Characteristics of each escitalopram study included.

doi:10.1371/journal.pone.0004824.t003

Our main results show that the pooled effect sizes for placebo response in escitalopram trials are 1.46 (95% CI 1.38 to 1.53) using the fixed-effects model and 1.48 (95% CI 1.26 to 1.69) using the random-effects model; and, for rTMS studies, the sham pooled effect size is, in the fixed-effects model, 0.77 (95% CI 0.66 to 0.88) and 0.82 (95% CI 0.63 to 1) in the random-effects model (Fig. 2). Since heterogeneity is significant in both analyses (χ2 = 86.54, p<0.001 and χ2 = 66.87, p<0.001, respectively) subsequent analyses were performed using the random-effects model. For both arms, sensitivity analysis and Begg's funnel plot show neither change in results after the exclusion of any particular study nor evidence of publication bias and systematic heterogeneity across the studies (Figures S1 and S2).

thumbnail

Figure 2. Forest plots showing placebo response in control groups of escitalopram (A) and rTMS (B) studies.

Forest plots show effect sizes from the random effects model. A negative effect indicates that endpoint depression scores in control groups are higher than baseline scores. Effect sizes are Cohen's d (standardized mean difference), error bars represent the 95% confidence interval.

doi:10.1371/journal.pone.0004824.g002

Subgroup analyses also show that the sham effect size of add-on rTMS studies (1.47, 95% CI 1.24–1.70) differ from studies not adopting such strategy (0.56, 95% CI 0.44–0.7, p<0.001), and of non-refractoriness studies (1.24, 95% CI 0.96–1.52) vs. studies with treatment-resistant patients (0.55, 95% CI 0.42 to 0.68, p<0.001) (Figure S3).

Exploratory analysis

Simple linear regressions (table 4) show that some variables are associated with the outcome: (1) a negative association is observed for treatment resistant patients (ß coefficient = −0.69, p<0.001), meaning that refractoriness diminish placebo response; (2) a positive association (ß coefficient = 0.9, p<0.001) is observed for rTMS as add-on therapy, meaning that placebo response increases in accelerating studies and; (3) a positive association is observed for depression improvement in active groups for both escitalopram (B = 0.7, p<0.0001) and rTMS studies (B = 0.33, p = 0.002), i.e., studies showing a large depression improvement in active group also showed a large depression improvement in the control group.

thumbnail

Table 4. Meta regression results in which several variables were analyzed trough simple linear regressions.

doi:10.1371/journal.pone.0004824.t004

On the contrary, baseline HDRS scores, baseline MADRS scores, depression scale utilized, gender and age are not associated with placebo response for both escitalopram and rTMS studies - except for baseline HDRS scores in rTMS studies (B = 0.52, p = 0.04), i.e., depression severity associates with a large placebo response.

Finally, variables that were associated with the outcome (p<0.1) were included in multiple linear regressions (Table 5). We observe that in models 1a (covariates: active-rTMS treatment and add-on therapy) and 1b (covariates: HDRS baseline scores and add-on therapy) only rTMS as add-on therapy remains associated with the outcome. Similarly, the variable baseline HDRS scores looses significance when meta-regressed together with treatment resistant (model 2b); however, in model 2a, both depression improvement in active-rTMS group (B = 0.2, p = 0.02) and treatment resistant (B = −0.57, p<0.0001) associates with placebo response. In model 3 all mentioned variables are regressed together; results show that only rTMS as add-on therapy still positively associates with the outcome (B = 0.53, p = 0.02), although there is still a trend for a negative association for treatment resistant variable (B = −0.31, p = 0.08), meaning that such variables still modify placebo response even when controlled by other significant variables.

thumbnail

Table 5. Exploratory regression models for rTMS studies in which significant results (obtained from simple linear regressions) were forced into several models.

doi:10.1371/journal.pone.0004824.t005

Discussion

This meta-analysis includes data from 12 escitalopram and 29 rTMS trials, assessing 2394 subjects in placebo/sham groups. Our main result shows that placebo response is large in major depression trials, regardless of the placebo method. Exploratory analyses found that patients with severe depression and with treatment-resistant depression present a lower placebo response; while in trials that rTMS is initiated concomitantly with an antidepressant drug, the placebo response is larger.

The main finding of our study is that both placebo interventions are associated with a large effect size in major depressive disorder, which is in line with previous studies: Walsh et al. [1] reviewed 75 depression trials and concluded that placebo response is substantial and increasing over years; Stein et al. [35] in a pooled analysis of five escitalopram trials showed that placebo response ranged from 31.6% to 45.9%; and Kirsch et al [2], reviewing 35 published and unpublished trials, showed that placebo response ranged from 0.7 to 1.1 Cohen's d. Therefore our study confirms that placebo response is substantial in pharmacological and non-pharmacological trials in major depression.

Another finding of our study is that placebo-drug response appears to be larger than sham-rTMS response – even controlling for treatment refractoriness. Even considering that the small difference might not be meaningful, this finding is contrary to conventional wisdom that sham devices would have a higher placebo response than placebo pills [8], [36]. In fact, a non-pill intervention showed increased response than a placebo pill in a prospective sham device vs. inert pill trial [37] and in a meta-analysis comparing subcutaneous placebo with oral placebo from acute migraine [38]. Our finding does however agree with a smaller acute care study that found no difference between parenteral medication and oral medication [39]. These differences could be related to the concept that placebo response is very heterogeneous and influenced by many variables. In our study, this finding might be explained by several factors:

  1. Study populations are different: 73% of rTMS trials enrolled refractory MDD patients; whereas no escitalopram trials enrolled refractory patients – in fact, STAR*D (Sequenced Treatment Alternatives to Relieve Depression) trial shows that remission rates decay at each time an antidepressant drug fail, being only 13% for refractory patients – i.e. patients who failed to remit after two trials [40]; and, since antidepressant drug effect is partially composed by a non-specific, placebo effect, placebo response might also decay in refractory patients. Along with these lines, low placebo responses were reported in a recent rTMS meta-analysis that addressed treatment-resistant patients [41] as well as two drug meta-analyses using lithium [42] and atypical antipsychotics [43].
  2. Study designs are different: although escitalopram and rTMS trials present comparable quality, they mainly differ in blinding quality, as adequate blinding is more difficult to obtain in non-pharmacological interventions [44]. The rTMS trials assessed used an approach in which patients and raters were blinded to the treatment group allocated; however, it is possible that rTMS appliers unconsciously behave different when applying real and sham stimulation as well as that patients discover in which intervention they were allocated. Unsuccessful blinding biases the results as expectation effects and intervention confidence will be lost [45], [46], [47], [48], therefore diminishing placebo response. Also, it is possible that study design influences outcome, since Woods et al. [49] showed that, in schizophrenia controlled-trials, improvement was larger in trials having no placebo arm; and Trivedi et al. [50] showed that response raters were different in depression controlled-trials regarding using or not a placebo run-in phase.
  3. Study sites and approaches are different: whereas drug trials are conducted along 8 weeks, with weekly returns, rTMS trials are conducted in 2 to 4 weeks – therefore longer exposure might be associated with a larger placebo response. On the other hand, rTMS treatment is associated with an intensive 10-day treatment (as opposed to weekly or bi-weekly interaction in drug trials) and this could potentiate placebo response in the rTMS trials.

Our results show that sham-response is smaller in trials that rTMS is not used as add-on therapy (0.56 vs. 1.24), suggesting that such device might not be associated with a large placebo effect, a finding that was also observed in meta-analyses of Parkinson's disease [51] and of refractory MDD [41]. Also, add-on rTMS trials improve response in placebo arm even when controlled for other variables, which could point to a synergistic effect between sham-rTMS and the drug, since there is no association between placebo response and previous use of antidepressant drugs. Finally, sham method (sham coil vs. angled coil) does not change placebo response – perhaps because both approaches, in fact, do not guarantee blinding.

Limitations

There was significant between-study heterogeneity in our meta-analysis, suggesting that the variation of effect size estimates in the studies were more than expected by chance. To address this limitation, we (1) used a random-effects model, which is a more conservative pooled analysis that take into account the between-study heterogeneity; (2) performed sensitivity analyses, to address whether the exclusion of an study could affect the pooled effect size; (3) assessed the quality of each study, looking for potential biases; and (4) checked for publication biases using Begg's funnel plot.

Another limitation is that, for pharmacological studies, we only included escitalopram studies; consequently, it is possible that the placebo response of other drugs is different. However, our study is in line with previous meta-analyses that showed similar placebo responses in major depression studies [1], [52], assessed a significant number (1714) of patients and included unpublished studies; therefore this hypothesis is less likely.

Finally, it should be emphasized that the secondary analyses performed are exploratory and might be underpowered; in fact, since ten linear regressions have been performed in each pooled analysis, there is a 50% probability of observing one positive association merely due to chance.

Clinical implications

Because we addressed the influence of several variables in sham-response, our results have some implications for future rTMS trial designs, such as: (1) sham device method is not associated with placebo response; therefore this factor seems less relevant than currently considered by the researchers in this field [53]; (2) age and gender are probably not related with placebo response - although age seems to be related to depression response in some studies [54], [55]; (3) refractoriness is associated with a lower placebo response – and, in fact, a lower depression response [41], [54] [56]; perhaps indicating that such patients are very unresponsive to any intervention at all and therefore rTMS studies should focus on non-refractory patients or, on the contrary, the positive results of rTMS trials might be due to a lower placebo response that increases active-sham difference – therefore, future rTMS trials should quantify the degree of refractoriness of each patient, and; (4) placebo response is high in add-on rTMS trials – this could indicate there is a synergistic effect with the drug and, therefore, future trials could use a two-way factorial design (i.e., sham vs. real-rTMS and placebo vs. active drug) to address the relationship among rTMS and drug interventions.

Our study also stresses the heterogeneity of placebo response in different contexts and interventions; therefore, the lower placebo response observed in sham trials could be explored by using a qualitative approach to understand patient's expectancies regarding rTMS intervention or, perhaps, by a sham-device vs. inert pill trial, in the same fashion of a prior placebo study [37].

Final remarks

In summary, our study shows that placebo response in rTMS and escitalopram trials is large and appears to be lower for rTMS trials. The sham response is negatively associated with refractoriness and positively associated with rTMS add-on studies; whereas sham method utilized, age and gender are not associated with a greater sham response. It is possible that design issues such as the lack of adequate blinding associate with lower placebo responses; however, we cannot measure in which extent such difference is explained by other cultural factors, as pill-taking healing is a mainstream medical ritual, while sham devices are not. The sham response of rTMS significantly varies among studies and can influence the results of a clinical trial as it will determine the effect size of a given sham-controlled trial, therefore, further studies are needed to explore its effects as to design appropriate sham-controlled randomized clinical trials.

Supporting Information

Table S1.

The file contains the quality assessment of all the studies included.

doi:10.1371/journal.pone.0004824.s001

(0.57 MB PPT)

Figure S1.

(A) shows the sensitivity analysis, assessing the individual influence of a particular study by showing the resulting effect size and 95% confidence interval (CI) after its exclusion. (B) shows the funnel plot of the effect sizes (Cohen's d) according to their standard errors. Cohen's d is the standardized mean difference, error bars represent the 95% CI.

doi:10.1371/journal.pone.0004824.s002

(3.00 MB TIF)

Figure S2.

(A) shows the sensitivity analysis, assessing the individual influence of a particular study by showing the resulting effect size and 95% confidence interval (CI) after its exclusion. (B) shows the funnel plot of the effect sizes (Cohen's d) according to their standard errors. Cohen's d is the standardized mean difference, error bars represent the 95% CI.

doi:10.1371/journal.pone.0004824.s003

(3.00 MB TIF)

Figure S3.

(A) shows the influence of the variable add-on rTMS in the pooled analysis of the studies, by pooling together only studies in which this variable is present (top) or absent (bottom) and thereby comparing the resulting effect sizes (Cohen's d, standardized mean difference). (B) shows the influence of the variable treatment-resistant depression, when it is present (top) or absent (bottom) in the resulting effect sizes.

doi:10.1371/journal.pone.0004824.s004

(3.00 MB TIF)

Acknowledgments

We wish to thank Forest Research Institute and all contacted authors that kindly provided us the data to perform the present study.

Author Contributions

Conceived and designed the experiments: FF. Analyzed the data: ARB ML. Wrote the paper: ARB ML TK FF.

References

  1. 1. Walsh BT, Seidman SN, Sysko R, Gould M (2002) Placebo response in studies of major depression: variable, substantial, and growing. Jama 287: 1840–1847.
  2. 2. Kirsch I, Deacon BJ, Huedo-Medina TB, Scoboria A, Moore TJ, et al. (2008) Initial severity and antidepressant benefits: a meta-analysis of data submitted to the Food and Drug Administration. PLoS Med 5: e45.
  3. 3. Martin JL, Barbanoj MJ, Schlaepfer TE, Thompson E, Perez V, et al. (2003) Repetitive transcranial magnetic stimulation for the treatment of depression. Systematic review and meta-analysis. Br J Psychiatry 182: 480–491.
  4. 4. Couturier JL (2005) Efficacy of rapid-rate repetitive transcranial magnetic stimulation in the treatment of depression: a systematic review and meta-analysis. J Psychiatry Neurosci 30: 83–90.
  5. 5. Gross M, Nakamura L, Pascual-Leone A, Fregni F (2007) Has repetitive transcranial magnetic stimulation (rTMS) treatment for depression improved? A systematic review and meta-analysis comparing the recent vs. the earlier rTMS studies. Acta Psychiatr Scand 116: 165–173.
  6. 6. Schutter DJ (2008) Antidepressant efficacy of high-frequency transcranial magnetic stimulation over the left dorsolateral prefrontal cortex in double-blind sham-controlled designs: a meta-analysis. Psychol Med 1–11.
  7. 7. Herrmann LL, Ebmeier KP (2006) Factors modifying the efficacy of transcranial magnetic stimulation in the treatment of depression: a review. J Clin Psychiatry 67: 1870–1876.
  8. 8. Kaptchuk TJ, Goldman P, Stone DA, Stason WB (2000) Do medical devices have enhanced placebo effects? J Clin Epidemiol 53: 786–792.
  9. 9. Posternak MA, Zimmerman M (2007) Therapeutic effect of follow-up assessments on antidepressant and placebo response rates in antidepressant efficacy trials: meta-analysis. Br J Psychiatry 190: 287–292.
  10. 10. Stolk P, Ten Berg MJ, Hemels ME, Einarson TR (2003) Meta-analysis of placebo rates in major depressive disorder trials. Ann Pharmacother 37: 1891–1899.
  11. 11. Burke WJ, Gergel I, Bose A (2002) Fixed-dose trial of the single isomer SSRI escitalopram in depressed outpatients. J Clin Psychiatry 63: 331–336.
  12. 12. Clayton AH, Croft HA, Horrigan JP, Wightman DS, Krishen A, et al. (2006) Bupropion extended release compared with escitalopram: effects on sexual functioning and antidepressant efficacy in 2 randomized, double-blind, placebo-controlled studies. J Clin Psychiatry 67: 736–746.
  13. 13. Rapaport MH, Bose A, Zheng H (2004) Escitalopram continuation treatment prevents relapse of depressive episodes. J Clin Psychiatry 65: 44–49.
  14. 14. Bose A, Li D, Gandhi C (2008) Escitalopram in the acute treatment of depressed patients aged 60 years or older. Am J Geriatr Psychiatry 16: 14–20.
  15. 15. Ninan P, Ventura D, Wang JEscitalopram is effective and well tolerated in the treatment of severe depression. 2003; San Francisco, USA.
  16. 16. Alexopoulos G, Gordon J, Zhang DA placebo-controlled trial of escitalopram and sertraline in the treatment of major depressive disorder. 2004; San Juan, Puerto Rico.
  17. 17. Kasper S, de Swart H, Friis Andersen H (2005) Escitalopram in the treatment of depressed elderly patients. Am J Geriatr Psychiatry 13: 884–891.
  18. 18. Nierenberg AA, Greist JH, Mallinckrodt CH, Prakash A, Sambunaris A, et al. (2007) Duloxetine versus escitalopram and placebo in the treatment of patients with major depressive disorder: onset of antidepressant action, a non-inferiority study. Curr Med Res Opin 23: 401–416.
  19. 19. Higgins J, Green S, Cochrane Handbook for Systematic Reviews of Interventions Version 5.0.0 (updated February 2008) The Cochrane Collaboration, 2008. Available from www.cochrane-handbook.org.
  20. 20. Poulet E, Brunelin J, Boeuve C, Lerond J, D'Amato T, et al. (2004) Repetitive transcranial magnetic stimulation does not potentiate antidepressant treatment. Eur Psychiatry 19: 382–383.
  21. 21. Herwig U, Lampe Y, Juengling FD, Wunderlich A, Walter H, et al. (2003) Add-on rTMS for treatment of depression: a pilot study using stereotaxic coil-navigation according to PET data. J Psychiatr Res 37: 267–275.
  22. 22. Rossini D, Lucca A, Zanardi R, Magri L, Smeraldi E (2005) Transcranial magnetic stimulation in treatment-resistant depressed patients: a double-blind, placebo-controlled trial. Psychiatry Res 137: 1–10.
  23. 23. Jorge RE, Moser DJ, Acion L, Robinson RG (2008) Treatment of vascular depression using repetitive transcranial magnetic stimulation. Arch Gen Psychiatry 65: 268–276.
  24. 24. Egger M, Davey Smith G, Schneider M, Minder C (1997) Bias in meta-analysis detected by a simple, graphical test. Bmj 315: 629–634.
  25. 25. Bretlau LG, Lunde M, Lindberg L, Unden M, Dissing S, et al. (2008) Repetitive transcranial magnetic stimulation (rTMS) in combination with escitalopram in patients with treatment-resistant major depression: a double-blind, randomised, sham-controlled trial. Pharmacopsychiatry 41: 41–47.
  26. 26. O'Reardon JP, Cristancho P, Pilania P, Bapatla KB, Chuai S, et al. (2007) Patients with a major depressive episode responding to treatment with repetitive transcranial magnetic stimulation (rTMS) are resistant to the effects of rapid tryptophan depletion. Depress Anxiety 24: 537–544.
  27. 27. Altshuler LL, Frye MA, Gitlin MJ (2003) Acceleration and augmentation strategies for treating bipolar depression. Biol Psychiatry 53: 691–700.
  28. 28. Hoppner J, Schulz M, Irmisch G, Mau R, Schlafke D, et al. (2003) Antidepressant efficacy of two different rTMS procedures. High frequency over left versus low frequency over right prefrontal cortex compared with sham stimulation. Eur Arch Psychiatry Clin Neurosci 253: 103–109.
  29. 29. Loo CK, Mitchell PB, McFarquhar TF, Malhi GS, Sachdev PS (2007) A sham-controlled trial of the efficacy and safety of twice-daily rTMS in major depression. Psychol Med 37: 341–349.
  30. 30. Herwig U, Fallgatter AJ, Hoppner J, Eschweiler GW, Kron M, et al. (2007) Antidepressant effects of augmentative transcranial magnetic stimulation: randomised multicentre trial. Br J Psychiatry 191: 441–448.
  31. 31. Fitzgerald PB, Benitez J, de Castella A, Daskalakis ZJ, Brown TL, et al. (2006) A randomized, controlled trial of sequential bilateral repetitive transcranial magnetic stimulation for treatment-resistant depression. Am J Psychiatry 163: 88–94.
  32. 32. Fitzgerald PB, Brown TL, Marston NA, Daskalakis ZJ, De Castella A, et al. (2003) Transcranial magnetic stimulation in the treatment of depression: a double-blind, placebo-controlled trial. Arch Gen Psychiatry 60: 1002–1008.
  33. 33. Loo CK, Mitchell PB, Croker VM, Malhi GS, Wen W, et al. (2003) Double-blind controlled investigation of bilateral prefrontal transcranial magnetic stimulation for the treatment of resistant major depression. Psychol Med 33: 33–40.
  34. 34. Anderson IM, Delvai NA, Ashim B, Ashim S, Lewin C, et al. (2007) Adjunctive fast repetitive transcranial magnetic stimulation in depression. Br J Psychiatry 190: 533–534.
  35. 35. Stein DJ, Baldwin DS, Dolberg OT, Despiegel N, Bandelow B (2006) Which factors predict placebo response in anxiety disorders and major depression? An analysis of placebo-controlled studies of escitalopram. J Clin Psychiatry 67: 1741–1746.
  36. 36. Walach H (2001) The efficacy paradox in randomized controlled trials of CAM and elsewhere: beware of the placebo trap. J Altern Complement Med 7: 213–218.
  37. 37. Kaptchuk TJ, Stason WB, Davis RB, Legedza AR, Schnyer RN, et al. (2006) Sham device v inert pill: randomised controlled trial of two placebo treatments. BMJ 332: 391–397.
  38. 38. de Craen AJ, Tijssen JG, de Gans J, Kleijnen J (2000) Placebo effect in the acute treatment of migraine: subcutaneous placebos are better than oral placebos. J Neurol 247: 183–188.
  39. 39. Schwartz NA, Turturro MA, Istvan DJ, Larkin GL (2000) Patients' perceptions of route of nonsteroidal anti-inflammatory drug administration and its effect on analgesia. Acad Emerg Med 7: 857–861.
  40. 40. Rush AJ, Trivedi MH, Wisniewski SR, Nierenberg AA, Stewart JW, et al. (2006) Acute and longer-term outcomes in depressed outpatients requiring one or several treatment steps: a STAR*D report. Am J Psychiatry 163: 1905–1917.
  41. 41. Lam RW, Chan P, Wilkins-Ho M, Yatham LN (2008) Repetitive transcranial magnetic stimulation for treatment-resistant depression: a systematic review and metaanalysis. Can J Psychiatry 53: 621–631.
  42. 42. Baethge C, Smolka MN, Gruschka P, Berghofer A, Schlattmann P, et al. (2003) Does prophylaxis-delay in bipolar disorder influence outcome? Results from a long-term study of 147 patients. Acta Psychiatr Scand 107: 260–267.
  43. 43. Papakostas GI, Shelton RC, Smith J, Fava M (2007) Augmentation of antidepressants with atypical antipsychotic medications for treatment-resistant major depressive disorder: a meta-analysis. J Clin Psychiatry 68: 826–831.
  44. 44. Boutron I, Guittet L, Estellat C, Moher D, Hrobjartsson A, et al. (2007) Reporting methods of blinding in randomized trials assessing nonpharmacological treatments. PLoS Med 4: e61.
  45. 45. Fergusson D, Glass KC, Waring D, Shapiro S (2004) Turning a blind eye: the success of blinding reported in a random sample of randomised, placebo controlled trials. Bmj 328: 432.
  46. 46. Hrobjartsson A (2002) What are the main methodological problems in the estimation of placebo effects? J Clin Epidemiol 55: 430–435.
  47. 47. Schulz KF, Chalmers I, Hayes RJ, Altman DG (1995) Empirical evidence of bias. Dimensions of methodological quality associated with estimates of treatment effects in controlled trials. Jama 273: 408–412.
  48. 48. Moher D, Pham B, Jones A, Cook DJ, Jadad AR, et al. (1998) Does quality of reports of randomised trials affect estimates of intervention efficacy reported in meta-analyses? Lancet 352: 609–613.
  49. 49. Woods SW, Gueorguieva RV, Baker CB, Makuch RW (2005) Control group bias in randomized atypical antipsychotic medication trials for schizophrenia. Arch Gen Psychiatry 62: 961–970.
  50. 50. Trivedi MH, Rush H (1994) Does a placebo run-in or a placebo treatment cell affect the efficacy of antidepressant medications? Neuropsychopharmacology 11: 33–43.
  51. 51. Fregni F, Simon DK, Wu A, Pascual-Leone A (2005) Non-invasive brain stimulation for Parkinson's disease: a systematic review and meta-analysis of the literature. J Neurol Neurosurg Psychiatry 76: 1614–1623.
  52. 52. Turner EH, Matthews AM, Linardatos E, Tell RA, Rosenthal R (2008) Selective publication of antidepressant trials and its influence on apparent efficacy. N Engl J Med 358: 252–260.
  53. 53. Loo CK, Taylor JL, Gandevia SC, McDarmont BN, Mitchell PB, et al. (2000) Transcranial magnetic stimulation (TMS) in controlled treatment studies: are some “sham” forms active? Biol Psychiatry 47: 325–331.
  54. 54. Fregni F, Marcolin MA, Myczkowski M, Amiaz R, Hasey G, et al. (2006) Predictors of antidepressant response in clinical trials of transcranial magnetic stimulation. Int J Neuropsychopharmacol 9: 641–654.
  55. 55. Mitchell PB, Loo CK (2006) Transcranial magnetic stimulation for depression. Aust N Z J Psychiatry 40: 406–413.
  56. 56. Lisanby SH, Husain MM, Rosenquist PB, Maixner D, Gutierrez R, et al. (2008) Daily Left Prefrontal Repetitive Transcranial Magnetic Stimulation in the Acute Treatment of Major Depression: Clinical Predictors of Outcome in a Multisite, Randomized Controlled Clinical Trial. Neuropsychopharmacology.
  57. 57. Boutros NN, Gueorguieva R, Hoffman RE, Oren DA, Feingold A, et al. (2002) Lack of a therapeutic effect of a 2-week sub-threshold transcranial magnetic stimulation course for treatment-resistant depression. Psychiatry Res 113: 245–254.
  58. 58. Koerselman F, Laman DM, van Duijn H, van Duijn MA, Willems MA (2004) A 3-month, follow-up, randomized, placebo-controlled study of repetitive transcranial magnetic stimulation in depression. J Clin Psychiatry 65: 1323–1328.
  59. 59. Holtzheimer PE 3rd, Russo J, Claypoole KH, Roy-Byrne P, Avery DH (2004) Shorter duration of depressive episode may predict response to repetitive transcranial magnetic stimulation. Depress Anxiety 19: 24–30.
  60. 60. Jorge RE, Robinson RG, Tateno A, Narushima K, Acion L, et al. (2004) Repetitive transcranial magnetic stimulation as treatment of poststroke depression: a preliminary study. Biol Psychiatry 55: 398–405.
  61. 61. Mosimann UP, Schmitt W, Greenberg BD, Kosel M, Muri RM, et al. (2004) Repetitive transcranial magnetic stimulation: a putative add-on treatment for major depression in elderly patients. Psychiatry Res 126: 123–133.
  62. 62. Hausmann A, Kemmler G, Walpoth M, Mechtcheriakov S, Kramer-Reinstadler K, et al. (2004) No benefit derived from repetitive transcranial magnetic stimulation in depression: a prospective, single centre, randomised, double blind, sham controlled “add on” trial. J Neurol Neurosurg Psychiatry 75: 320–322.
  63. 63. Rossini D, Magri L, Lucca A, Giordani S, Smeraldi E, et al. (2005) Does rTMS hasten the response to escitalopram, sertraline, or venlafaxine in patients with major depressive disorder? A double-blind, randomized, sham-controlled trial. J Clin Psychiatry 66: 1569–1575.
  64. 64. Rumi DO, Gattaz WF, Rigonatti SP, Rosa MA, Fregni F, et al. (2005) Transcranial magnetic stimulation accelerates the antidepressant effect of amitriptyline in severe depression: a double-blind placebo-controlled study. Biol Psychiatry 57: 162–166.
  65. 65. Su TP, Huang CC, Wei IH (2005) Add-on rTMS for medication-resistant depression: a randomized, double-blind, sham-controlled trial in Chinese patients. J Clin Psychiatry 66: 930–937.
  66. 66. Januel D, Dumortier G, Verdon CM, Stamatiadis L, Saba G, et al. (2006) A double-blind sham controlled study of right prefrontal repetitive transcranial magnetic stimulation (rTMS): therapeutic and cognitive effect in medication free unipolar depression during 4 weeks. Prog Neuropsychopharmacol Biol Psychiatry 30: 126–130.
  67. 67. Avery DH, Holtzheimer PE 3rd, Fawaz W, Russo J, Neumaier J, et al. (2006) A controlled study of repetitive transcranial magnetic stimulation in medication-resistant major depression. Biol Psychiatry 59: 187–194.
  68. 68. Garcia-Toro M, Salva J, Daumal J, Andres J, Romera M, et al. (2006) High (20-Hz) and low (1-Hz) frequency transcranial magnetic stimulation as adjuvant treatment in medication-resistant depression. Psychiatry Res 146: 53–57.
  69. 69. Stern WM, Tormos JM, Press DZ, Pearlman C, Pascual-Leone A (2007) Antidepressant effects of high and low frequency repetitive transcranial magnetic stimulation to the dorsolateral prefrontal cortex: a double-blind, randomized, placebo-controlled trial. J Neuropsychiatry Clin Neurosci 19: 179–186.
  70. 70. Bortolomasi M, Minelli A, Fuggetta G, Perini M, Comencini S, et al. (2007) Long-lasting effects of high frequency repetitive transcranial magnetic stimulation in major depressed patients. Psychiatry Res 150: 181–186.
  71. 71. Mogg A, Pluck G, Eranti SV, Landau S, Purvis R, et al. (2008) A randomized controlled trial with 4-month follow-up of adjunctive repetitive transcranial magnetic stimulation of the left prefrontal cortex for depression. Psychol Med 38: 323–333.
  72. 72. Wade A, Michael Lemming O, Bang Hedegaard K (2002) Escitalopram 10 mg/day is effective and well tolerated in a placebo-controlled study in depression in primary care. Int Clin Psychopharmacol 17: 95–102.
  73. 73. Lepola UM, Loft H, Reines EH (2003) Escitalopram (10–20 mg/day) is effective and well tolerated in a placebo-controlled study in depression in primary care. Int Clin Psychopharmacol 18: 211–217.
  74. 74. Wagner KD, Jonas J, Findling RL, Ventura D, Saikali K (2006) A double-blind, randomized, placebo-controlled trial of escitalopram in the treatment of pediatric depression. J Am Acad Child Adolesc Psychiatry 45: 280–288.