Placebo Effect in Obsessive-Compulsive Disorder (OCD). Placebo Response and Placebo Responders in OCD: The Trend Over Time.

BACKGROUND: Placebo response appears to be increasing in antidepressant, antipsychotic and various internal medicine trials. A similar trend has been reported for OCD during 1989-1999. Placebo response is generally considered as the extent to which placebo treatment is associated with core symptom improvement. In this analysis, we used Joinpoint regression to assess the time trend of both placebo response and placebo responder rates according to the year of publication with no time restriction in OCD drug trials.
METHODS: We included drug and/or psychotherapy trials vs. placebo from PubMed, Embase, CINAHL, and PsycINFO retrieved through the search (placebo OR sham) AND (obsessive* OR OCD). We included studies through investigator consensus. We then performed on data of included studies log-linear joinpoint segmented regression models using a p<0.05 cutoff.
RESULTS: We included 113 studies from 112 published papers. Placebo mean annual response rates in OCD studies significantly increased from 1991 to 2017 with an annual percent change (APC) of 0.66%, while placebo mean annual responder rates also significantly increased from 2010 to 2017, with an APC of 5.45%. Drug mean annual response rates in OCD studies significantly increased from 1987 to 2012 with an APC of 0.72%, while the corresponding responder rates did not show statistically significant APC changes between 1984 and 2017.
CONCLUSION: We observed a tendency for placebo to increase both measures of response in OCD clinical drug trials through the years that tend to approximate the responses shown by drugs. Changes in the type of study (moving from classical head to head comparisons to add-on studies in treatmentresistant populations) and countries involved in experimentation may partially account for some portion of these results. It appears that placebo effects are becoming more elusive and out of control. Copyright© Bentham Science Publishers; For any queries, please email at epub@benthamscience.net.

INTRODUCTION

A common belief lasting until the early nineties was that obsessive-compulsive disorder (OCD) did not respond to placebo [1]. However, this view rapidly changed as reliable scales developed to measure OCD symptoms.

There are two ways to measure the response of a patient or of a group of patients to a given treatment, drug or placebo, i.e., to consider improvement with respect to a baseline on a predetermined rating scale in terms of points or percentage, hence providing a measure of the percent response of a group, and to define some criteria for responsiveness and provide the percentage of patients who reached or surpassed a given threshold to qualify as treatment responders. A further specification of the latter is the remitter status, based on even stricter criteria. We will call the former “placebo (or drug) effect” to differentiate it from the ambiguous term “placebo (or drug) response”, and term the latter responder rate (to placebo or drug) so to hold the two concepts apart. In OCD, the most commonly used rating scale as a primary outcome measure is the Yale-Brown Obsessive-Compulsive Scale (Y-BOCS) [2, 3] and responses to treatments, or treatment effects, as we term them to avoid confusion with number of persons who are considered as treatment responders, are rated as percent variations of scores on the Y-BOCS from baseline at a given time-point. On the other hand, treatment responder rates are considered as the fraction of patients in the sample who are judged on the basis of established criteria, usually an at least 35% (or 25%, according to the study’s author choice) drop of Y-BOCS scores from baseline or 1 or 2 (very much or much improved, respectively) on the Clinical Global Impressions scale, improvement version [4].

There is evidence for the growth of placebo effect (and responder rates) in clinical trials across the years for depression [5] and, less consistently, bipolar disorder [6], although some biases may have influenced the conclusions of this second study [7]. Other psychiatric disorders have not been investigated for this year-of-publication effect specifically, but a multivariate meta-analysis carried-out by Ackerman and Greenland [8] found an increase in the effect of placebo through the years, confirming the impressions of others [9]. This could be due to the host of factors, like trends in the type of patient recruited (regarding illness duration, severity, and comorbidity), type of drug used (the side effect profile of a drug like clomipramine increases the likelihood of drug identification by both clinician and patient and may affect outcome), study site characteristics (that may reflect the characteristics of participating physicians and principal investigators and contribute to large across-sites differences which are usually disregarded in most reports), outcome measures employed, publication bias-file drawer effect, and last, but not least, the fact that the more effective the drug in a study, the more effective the placebo [10]. In fact, a publication year effect has been shown for OCD treatment across the years [11]. However, Ackerman and Greenland [8], who used meta-regression to evaluate placebo-controlled drug trials in OCD, did not include in their remarkable paper, drug trials in paediatric populations did not consider surgical procedures versus sham surgery or psychotherapies versus sham psychological interventions, and failed to consider a considerable number of papers covering a quite long period of time. In fact, they analysed a restricted period of time of placebo-controlled drug trials of three SSRIs and clomipramine in a period spanning from 1989 to 1999 (i.e., after the introduction of the Yale-Brown Obsessive Compulsive Scale, Y-BOCS), with a consequent loss of more than ten years of literature.

Systematic investigations of treatment of OCD started at the dawn of the eighties, with the use of clomipramine [12, 13]; clomipramine [14] and imipramine [15, 16] dominated the scene during the mid-eighties, and it was only during the late eighties that SSRIs, primarily fluvoxamine, were introduced [17, 18]. The first two studies comparing sertraline to placebo appeared in 1990 and yielded contrasting results [19, 20]. The first published trials of fluoxetine versus placebo appeared in 1992 [21], but regarded data that started being gathered in the late eighties [22-24], therefore simultaneous with, if not preceding those of sertraline. It is noteworthy that fluoxetine had received extensive open trials in OCD since 1985 [25], whereas for sertraline, the two aforementioned double-blind studies were the first studies of sertraline in OCD to be published [19, 20]. This publication lag may create a bias in the attempt to clarify whether the effect of a given treatment increased or decreased with time. Unfortunately, most studies do not provide the period during which they were conducted and render it difficult to correct for this bias. Hence, we will consider publication data as a factor despite realising that it does not exactly reflect the period during which the study has been carried out.

Our aim was to extend Ackerman and Greenland’s [8] observations beyond 2002, including also studies that did not use drugs, but other methods as well that could ensure double-blinding. We did not use the same method, but rather a JoinPoint regression.

METHODS

We carried-out a general search in the PubMed-MedLine-Index Medicus and Embase-Excerpta Medica and PsycLit-Psychological Abstracts databases using the following strategy: (placebo OR sham) AND (obsessive* OR OCD) with no time, language or any other restriction, but animal studies were subsequently excluded. We did not use the PubMed “Animal studies” function to exclude such studies, because such function often produces unreliable results. Papers were individually searched for adherence to our inclusion criteria. Retrieved relevant papers, comprising reviews and meta-analyses, were searched in their reference lists for providing additional papers with adequate research data. Final inclusion criteria for data analysis comprised: single or double-blind design, clearly stated assessment of response (responder rate or percent response on rating scales), sufficient time of treatment administration for the expected response to be observed, absence or adequate addressing of confounders that could render response not attributable to specific treatments. Specifically, the second part of cross-over studies was discarded if switching from one treatment to another had not a sufficient treatment-free wash-out period to avoid carry-over effects; survival studies were excluded when tapering-off of a combined therapy involved a drug vs. placebo when another drug or treatment was continued; add-ons were given not to patients stabilised on a given drug, but on drug-free or drug naïve populations. In this first report on placebo response in OCD we concentrate on double-blind studies using drugs, hence we excluded studies with psychotherapy or comparing mechanical devices with sham, like deep brain stimulation, electroconvulsive therapy, or deep/repetitive transcranial magnetic stimulation, provided they did not have a placebo and a drug arm. Excluded were also studies focusing on other than clinical outcomes, those carried-out on mixed populations (e.g., OCD and Tourette’s) without providing results specific for each subpopulation, and those with designs such that a placebo effect could not be calculated.

Statistical Analysis

We analysed temporal trends of placebo and drug response rates/responder proportions through log-linear joinpoint segmented regression models, which identify points corresponding to statistically significant changes over time in the linear slope of the occurring trend [26]. We used annual mean rates of placebo and drug effect (mean placebo and drug-induced improvements and mean placebo/drug responders) as independent variable assuming constant variance (homoscedasticity) without log transformation. We applied a grid search method to fit regression functions with unknown joinpoints assuming a Poisson distribution and uncorrelated errors. We set the minimum/maximum joinpoint number from 0 to 2, and used a permutation test with overall significance level set at p<0.05 and number of randomly permuted datasets of 4,499 to select the best fit. In the final model, each joinpoint indicates a trend change. We reported the estimated annual percent change (APC) for segmented analysis. Joinpoint analyses were performed using the Joinpoint Regression Program, version 3.5, from the US National Cancer Institute (https://surveillance.cancer.gov/joinpoint/).

RESULTS

Our PubMed-MedLine-Index Medicus and Embase-Excerpta Medica, CINAHL, and PsycLit-Psychological Abstracts searches yielded 886, 2932, 164, and 994 papers, respectively, as of April 24, 2018. The total output of our research is shown in Fig. (), which shows also the reasons for exclusion. All studies were searched for possible further includible papers. Included were 113 studies from 112 papers,

which met criteria for inclusion. The results of the included studies [13-15, 17, 18, 20-24, 28-128] are summarised in Table .

Joinpoint regression analyses of the period 1979-2017 showed that placebo mean annual effect rates in OCD studies significantly increased (APC value significantly differing from zero to α = 0.05 level) from 1991 to 2017 with an APC of 0.66% (p=0.04) following a period without statistically significant APC changes Fig. (). Placebo mean annual responder rates also significantly increased from 2010 to 2017 with an APC of 5.45% (p=0.02) following a period without statistically significant APC changes Fig. (). Drug mean annual effect rates in OCD studies significantly increased from 1987 to 2012 with an APC of 0.72% (p=0.04) between two periods without statistically significant APC changes Fig. (). Drug mean annual responder rates did not

show statistically significant APC changes between 1984 and 2017 Fig. ().

DISCUSSION

In this study, using Joinpoint regression analyses, we found significant increases in the net effect of placebo from 1991 to 2017 and in placebo responder rates from 2010 to 2017. The effects of drug treatment of OCD increased from 1987 to 2012, between two periods of no significant changes (1979-1987 and 2012-2017), while responder rates to drug treatment showed no statistically significant annual changes between 1984 and 2017. We confirmed the finding of a trend towards an increase of placebo effects in OCD drug treatment studies, previously reported by Ackerman and Greenland for the decade 1989 to 1999 in their 2002 meta-regression [8]. Surprisingly, we did not find drugs to increase their effect of responder rates in parallel with placebo. In fact, the increasing placebo effect that was found for other psychiatric (e.g., schizophrenia [129] and depression [130]) and non-psychiatric disorders (e.g., hypertension [131]), and placebo efficacy dragged drug efficacy to higher levels [130]. Here we found a trend towards increased placebo effects and responder rate, with a decrease in the difference between drug and placebo, to the point that the most recent response rates (effects) between placebo and drugs nearly overlap Figs. () and 4), while responder rates differ little Figs. () and 5); something similar has been described for schizophrenia [132], but not for depression [130]. Furthermore, we also found that effect/responder rates for OCD to both placebo and drugs are low compared to other psychiatric disorders, as recently reported using effect sizes as outcome measures [133]. Why OCD should be stiffer than other psychiatric disorders in responding to treatment may have a response to the pathophysiology of the disorder and to its related personality characteristics. That you can’t teach an old dog new tricks may well apply to this disorder. A therapeutic response may be hampered by too often controlling one’s health state, that prevents a patient from establishing an adequate clinical progress, and people with OCD often display pathological doubt, that prompts them to control for any change all too often and then to doubt for results.

It is not easy to explain the increase with time of placebo effects and responder rates we found here. One explanation that has been offered for similar results in depression is the “baseline inflation” [134], i.e., the tendency to inflate baseline scores of the scale chosen for subject inclusion in a randomised clinical trial (RCT) so to ensure more participants to the sponsor. This, combined with patient expectation and the Hawthorne-like effect of being closely observed, yields better results for drug and placebo alike. However, there has been no increase in baseline Y-BOCS scores of included samples.

Other possible explanations may involve historical factors. In fact, most early studies involved testing the efficacy of clomipramine and SSRIs in patients with OCD, or were survival studies focusing on Kaplan-Meyer curves after switching patients with a benefit on an antidepressant agent to placebo, measuring recurrence/relapse rates. In contrast, later studies increasingly focused on treatment-resistant populations and add-on drugs vs. placebo. It is highly probable that such populations are more resistant to the effects of both drugs and placebo. However, this should have been followed by a decrease in overall responsiveness, which we did not find; on the contrary, both effect and responder rates increased in later years, more so regarding placebo Fig. ()-5).

Another issue may regard the principal sites involved in the various studies. In antipsychotic drug trials, the increase in placebo response has been observed for North America-based studies, but not for those conducted in the rest of the world or for international studies including US sites [135], and the same phenomenon has been observed for painkiller trials, with sample size and study duration driving the placebo response increase [136]. Sample size in US studies correlated weakly with placebo response in our study (Pearson’s r=0.21). However, treatment duration did correlate strongly (Pearson’s r=0.54). Here we observed a curious phenomenon, i.e., that in studies 1980-2008, US-based studies prevailed over the rest of the world (N=52 vs. N=24), whereas in studies conducted from 2009 on, the rest of the world studies reversed the ratio (N=13, USA vs. N=24, rest of the world). The reversal was driven by Iran (N=17), a country that was not present during the 1980-2009 period. In Iranian studies, the correlation between sample size was much weaker than in US-based studies (r=0.093), perhaps a consequence of the fact that sample sizes in these studies were about 20 each with a much lower standard deviation than in US studies (Table ). In contrast with US studies, there was a strong negative correlation between treatment duration and response to placebo (r=-0.471). Despite the entity of placebo effect did not differ between US-based and Iran-based studies (Student’s t=1.145; p=0.256, not significant), we feel that the recent upsurge of placebo responder rates and the constant increase of placebo response are linked to the results of Iranian studies. The samples in American studies varied widely, as did the number of sites, while the Iranian studies recruited middle samples and tended to be single-center (Table ). It has been suggested in antidepressant trials for major depression that two factors that may be linked to reduced ability to detect a signal for an antidepressant are constituted by extremely large and extremely small samples and by multicentricity [137]; a reduced signal is usually linked to increasing placebo response that dampens the drug-placebo difference.

A possibility with longer-term treatment to associate with placebo response could reside in the exacerbating-remitting course that often characterizes OCD [138], especially in paediatric cases [139]. If the interpretation of the placebo effect as a regression to the mean holds true [140], it would ensure that by treating people for more time, there will be an increased probability of spontaneous remission of the disorder, that would be subsequently attributed to placebo. This matches the results of American studies, but is opposite to what Iranian studies tell us.

Still another possibility is a change in the characteristics of included patients. It has been speculated that RCTs tend to include patients repeatedly the same persons who participated in prior antidepressant trials, and this is usually addressed with excluding patients having participated recently in another RCT or having received psychotherapy in recent times. In fact, it was shown that patients with low income may be eager to participate in more than one antidepressant drug trial [141]. These patients would be expected to display a rather uniform behaviour in responding to treatment, thus favouring placebo response and their data tend to be increasingly included in databases, thus affecting results. In OCD, we do not have data at this regard, but the recent change in the classification of OCD spectrum disorders in the DSM-5 [142], which were previously classified amidst the anxiety disorders and included hoarding disorder [143], may have impacted the response of OCD to placebo. However, in this case, we should have expected a join point to occur about 2013, the year of introduction of DSM-5 [142]. In placebo effect, there was no such join point. Placebo effect showed a continuous growth from 1991 onwards, while placebo responder rate had a joinpoint at 2010, when the shift from DSM-IV-TR™ had still to occur. It should be said that DSM-IV-TR™ diagnoses in drug trials continued to be adopted along with DSM-5 diagnoses for some time. At any rate, it appears that in OCD, as in other mental and non-mental disorders, placebo effect and responder rates are puzzlingly increasing and appear to be out of control, and this depends on multiple factors [137, 144], pointing to changing populations included in RCTs (independently from initial severity) and prompting to a revision of the RCT model. Given that this trend is not specifically bound to a single condition, it is possible that it reflects continuing human evolution.

Limitations and Strengths

Our study is not a meta-analysis and the overall placebo effect and placebo-responder rates are not weighted for sample size. Furthermore, we did not include studies not using drugs, i.e., somatic treatments vs. sham, and did not distinguish between adult and paediatric studies. Moreover, we did not select our studies based on their quality nor did we address possible sponsor bias. However, our study is the first considering such a wide time period and the first to consider both net placebo effects and responder rates based on clear-cut criteria. Future studies will have to address the above concerns. It has been suggested that mechanical devices [145] and surgery [146] are endowed with a superior placebo effect than drugs, although the evidence is still inconclusive [147], and placebo, despite displaying large effects in depression, was not superior in non-pharmacological than in pharmacological studies in a meta-analysis [148]. Comparison between drug and somatic treatment of OCD will show whether in this disorder there is a strong mechanical device component in placebo response and whether there is an increase of response with the year of publication with these treatments similar to what occurs with drugs.

The fact that the curves of effect and responder rates did not reciprocally correspond for both placebo and drugs may be explained by the fact that some studies did not report one of them (Table ). Investigators need to report their data more clearly in the future, so to allow other investigators to perform meta-analyses on their data.

The current situation with OCD treatment is that this disorder is either treated with drugs having the ability to block the reuptake of serotonin or with cognitive-behavioral therapy or both, or with attempts to add-on ongoing pharmacotherapy other pharmacological agents having antidopaminergic or antiglutamatergic properties or somatic treatments like deep brain stimulation or transcranial magnetic stimulation. The first studies focused on the efficacy of drugs used classical designs and were carried-out by prestigious institutions. Once the concept that first line treatments were represented by SSRIs/clomipramine, the treatment paradigms shifted towards add-on and somatic treatments, and this may have affected the figures we obtained. The closing gap between placebo and drug treatment must prompt investigators to formulate new hypotheses to test and industries to produce alternatively working drugs.

CONCLUSION

In this Joinpoint regression analysis, we observed an increase of the response to placebo (placebo effect) as well as an increase in responder rates in OCD studies with the year of publication. Changes in study types and sites are apparently related to the results obtained. The gap between response to drug and response to placebo appears to be reducing to an extent that current therapeutic approaches to OCD are becoming questionable and should prompt to seek newer approaches in facing this stubborn disorder.

Table 1

Placebo and drug responsiveness in OCD studies.

Author(s)			Year						Affiliation/ Country				Population (N)											Design								Outcome Measures (Response Criteria)													Placebo Effect (%)									Placebo Responders (%)								Drug(s) or other Treatment than Placebo							Drug Effect (%)				Drug Responders (%)
Thorén et al. [13]			1980						Karolinska Universitetssjukhuset, Karolinska Institutet, Stockholm, Sweden				24; 8 placebo vs. 8 clomi vs. 8 nortriptyline											Clomi vs. nortriptyline vs. placebo × 5 wk								↓ from BL of OCD Scale derived from the CPRSResponders classified according to clinicians’ ratings													7% OCD Scale									Not given								Clomi 50→150 mg/dayNortriptyline 50→150 mg/day							42%;21% OCD Scale				54.54% including open clomi trial; Nortriptyline: Not given
Flament et al. [27]			1985						Child Psychiatry Branch, National Institute of Mental Health, Bethesda, MD, USA				19 paediatric patients; 10 received placebo first, 9 clomi											11-wk Randomised Cross-over (at wk 5) Trial; Clomi vs. placebo								↓O-C Rating Scale; no response criteria													10.37% OCR Scale																	Chlomi 50→max200 mg/day							32.6% OCR Scale
Mavissakalian et al. [14]			1985						Department of Psychiatry, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA				12; 5 placebo vs. 7 clomi											RCT clomi vs. placebo ×12 wk								Response: ↓Obsessive-Compulsive Neurotic Scale; Responders: Clinician Rating for OCD (5 points; score 1 on at least 3 points)													43% O-C Neurotic Scale									15.53% Clinician Rating for OCD								Clomi 50→max300 mg/day							35.13% O-C Neurotic Scale				43% Clinician Rating for OCD
Foa et al. [15]			1987						Department of Psychiatry, Medical College of Pennsylvania, Philadelphia, PA, USA				37, 18 (7 scoring high [≥21] on the BDI) placebo vs.19 (9 scoring high on the BDI) imipramine											DB RCT imipramine vs. placebo × 6 wk								Effect: ↓MOCI from BL													3.67%MOCI																	Imipramine 25→max250 mg/day							7.42% MOCI
Perse et al. [17]			1987						Anxiety Disorders Center, Department of Psychiatry, University of Wisconsin, Madison, WI, USA				20 randomised to placebo vs. flexible fluvoxamine doses; 4 drop-outs for various reasons left 16 patients available for the analysis; 8 placebo first, 8 fluvoxamine first											Fluvoxamine vs. placebo DB cross-over trial; placebo run-in × 2 wk → DB fluvoxamine vs. placebo × 8 wk → × 2 wk placebo → × 8 wk DB cross-over								↓from BL of Maudsley OC Inventory scores for treatment effect; Responder rate: Clinician’s judgement for response													-7% (Maudsley scores ↑)									19%								Fluvoxamine 50 → max 150 mg/day							14.54%				81%
Pato et al. [28]			1988						Laboratory of Clinical Science, NIMH, Bethesda, MD, USA				21 responders to clomi ×4 months; 71.42% had significant depression; all received placebo substitution											Clomi substituted by placebo in four days, then placebo ×7 wk; survival study (relapse rates)								Substitution effect: ↑Y-BOCS; ↑CPRS O-C; ↑NIMH-OC; Response: lack of relapse/recurrence; evidence of recurrence: development of “significant symptoms”													43.7% ↑Y-BOCS; 52.78% ↑CPRS O-C; 43.87% NIMH-OC									9.52% (OCD symptoms)								Clomi tapering-off (four days, half the dosage first, than all drug substituted with placebo)
Author(s)			Year						Affiliation/ Country						Population (N)									Design											Outcome Measures (Response Criteria)										Placebo Effect (%)									Placebo Responders (%)								Drug(s) or other Treatment than Placebo							Drug Effect (%)				Drug Responders (%)
Goodman et al. [18]			1989						Department of Psychiatry, Yale University School of Medicine and Connecticut Mental Health Center, Ribicoff Research Facilities, New Haven, CT, USA						42; 21 placebo vs. 21 fluvoxamine									Multicentre (2 sites), DB RCT fluvoxamine vs. placebo ×6-8 wk											Effect: ↓Y-BOCS score from BL; Response: CGIi 1-2 (used different scale, but rating is similar)										0% Y-BOCS									0%								Fluvoxamine started at 50 mg/day → max 300 mg/day							22.4% Y-BOCS				42.8%
Jenike et al. [29]			1989						Harvard Medical School and Massachusetts General Hospital, Boston, MA, USA						27: 14 placebo vs. 13 clomi									Multicentre (? sites), DB clomi vs. placebo × 10 wk											↓Y-BOCS scores from BL as treatment effect; no criterion for response, but stratification according to percentages of ↓Y-BOCS										8.46% ↓Y-BOCS									20-39% ↓Y-BOCS: 21.42%; ≥40% ↓Y-BOCS: 0%								Clomi 50→200 mg/day →max300 mg/day							36.93% ↓Y-BOCS				20-39% ↓Y-BOCS: 76.92%; ≥40% ↓Y-BOCS: 46.15%
Chouinard et al. [20]			1990						McGill University, Montréal, Québec, Canada						87, 44 placebo vs. 43 sertraline									Multicentre RCT DB sertraline, flexible doses vs. placebo × 10 wk											Treatment effect: % ↓Y-BOCS score from BL; % ↓NIMH-OC Scale score from BL; Response: CGIi 1-2										6.55% Y-BOCS; 6.13% NIMH-OC									11.364% CGIi								Sertraline 50→200 mg/day							16.2% Y-BOCS; 15.2% NIMH-OC				25.581% CGIi
Greist et al. [30]			1990						Department of Psychiatry, University of Wisconsin, Madison, WI, USA						31; 16 placebo vs. 15 clomi									Single site part of multicentre (21 sites) study, DB parallel RCT of clomi fixed→ flexible dose vs. placebo ×10 wk											Effect: ↓NIMH-OC and ↓Y-BOCS scores from BL; Response: Patient and Physician Global Evaluation (conceptually similar to CGIi 1-2)										6.97% ↓Y-BOCS;8.64% ↓NIMH-OC									5% “CGIi”								Clomi 25→200 max 300 mg/day							34.883% ↓Y-BOCS 27.912% ↓NIMH-OC				45% CGIi
Jenike et al. [31]			1990						Harvard Medical School, Boston, and the OCD Clinic and Research Unit, the Department of Molecular Biology, and the Inpatient Psychiatric Service, Massachusetts General Hospital, Boston, MA, USA						38; 20 placebo vs. 18 fluvoxamine									DB fluvoxamine vs. placebo × 10 wk											↓Y-BOCS scores from BL as treatment effect; no criterion for response										4% ↓Y-BOCS																	Fluvoxamine 50→max300 mg/day							16.81% ↓Y-BOCS
Author(s)			Year						Affiliation/ Country						Population (N)					Design															Outcome Measures (Response Criteria)										Placebo Effect (%)									Placebo Responders (%)								Drug(s) or other Treatment than Placebo							Drug Effect (%)				Drug Responders (%)
Katz et al. [32]			1990						New Drug Development Department-CNS Section, Pharmaceuticals Division, CIBA-GEIGY Corporation, Summit, NJ, USA						266 with Ham-D <17; 129 placebo vs. 134 clomi					Multicentre (12 sites), DB RCT clomi vs. placebo in pts. with DSM-III OCD and Ham-D <17 or Ham-D 17-21 (subsequently excluded from the analysis due to small sample size) ×10 wk →responders (CGIi ≤3) with treatment confirmed (placebo, N=12, clomi, N=101) ×42 wk															Effect: ↓NIMH-OC scores from BL;response: CGIi 1-2										1.96% at week 10;21.56% at week 52; 24.71% at end-point; NIMH-OC									9.3% 10 wk; 2.4% wk 52; 16.7% at end-point (wk 70) CGIi								Clomi 20→min 100, target 250, max 300 mg/day							34.69% at week 10; 55.1% at week 52; 46.94% at end-point, NIMH-OC				75.37% 10 weeks; 61.2% week 52; 72.2% at end-point (week 70) CGIi
Mavissakalian et al. [33]			1990						Department of Psychiatry, Ohio State University, Columbus, OH, USA						25; 12 placebo vs. 13 clomi					DB RCT clomi vs. placebo ×10 wk															Effect: ↓CY-BOCS from BL;response: much improved										1.11% Y-BOCS									0% Much improved								Clomi 50→2-weeks 200 mg/day →flexible							54.19% Y-BOCS				35% Much improved
Montgomery et al. [34]			1990						Imperial College London, St Mary's Hospital Medical School, London UK						14; 7 placebo first, 7 clomi first					4-wk DB clomi vs. placebo→cross-over ×4 wk															Criteria for effect or response not specified; assessment with 6-item obsessional scale extracted from CPRS and MADRS										5.2%(↓from BL of CPRS Obs Scale scores)																	Clomi 75 mg/day							64.5%(↓from BL of CPRS Obs Scale scores)
McDougle et al. [35]			1991						Connecticut Mental Health Center, New Haven, CT, USA						30 DR (failure to reach ≥35% ↓ from BL Y-BOCS scores and CGIi>2 and clinician consensus after fluvoxamine × 8 wk) at flexible doses up to 300 mg/day; 9 placebo vs. 11 add-on lithium [study 1]; 5 vs. 5 add-on lithium					DB RCT to add-on lithium vs. placebo ×2 wk to unchanged fluvoxamine (up to 300 mg/day) [study 1, 20 patients]; 4 wk blind placebo followed by 4 wk open lithium [study 2, 10 inpatients]															Y-BOCS (response: ↓≥35% drop from BL and Y-BOCS<16);CGIi 1-2; clinician consensus: all three=marked, 2 of 3=partial, <2 no response)										Study 1: -5.09% (Y-BOCS scores↑)Study 2: 5.83% (Y-BOCS)									Studies 1 & 2: 0% marked and partial (Y-BOCS, CGIi, clinician cons.)								Clomi 200-300 mg/day + lithium 900 mg/day →0.5-1.2 mEq in plasma							Study 1: 10.52% (Y-BOCS);Study 2:-10.66% (Y-BOCS scores↑)				Study 1: 9.091% marked, 9.091% partial (Y-BOCS, CGIi, clinician’s consensus); Study 2: 0% marked and partial
Author(s)			Year						Affiliation/ Country						Population (N)					Design															Outcome Measures (Response Criteria)										Placebo Effect (%)									Placebo Responders (%)								Drug(s) or other Treatment than Placebo							Drug Effect (%)				Drug Responders (%)
The Clomi Collaborative Study Group [36]			1991						Clinical Neuropharmacology, Glaxo Inc, Research Triangle Park, NC, USA						520; 120 placebo vs. 118 clomi (study 1)129 placebo vs. 134 clomi (study 2)					Multicentre (9 centres study 1; 12 centres, study 2), clomi vs. placebo × 10 wk; two identical studies															Effect: ↓CY-BOCS from BL; Response ↓≥35% from BL										3% Y-BOCSstudy 15% Y-BOCS study 2									7.5% Y-BOCS study 17% Y-BOCS study 2								Clomi 25→max300 mg/day							38% Y-BOCS study 144% Y-BOCS study 2				51% Y-BOCS study 160% Y-BOCS study 2
DeVeaugh-Geiss et al. [37]			1992						Clinical Neuropharmacology, Glaxo Inc, Research Triangle Park, NC, USA						60 children or adolescents 10-17 years, 29 placebo vs. 31 clomi					Multicentre (5 sites) DB RCT clomi vs. placebo × 8 wk															Effect: ↓Y-BOCS from BL;Response: CGIi 1-2										8% Y-BOCS									17% CGIi								Clomi 25→100, max 200 mg/day							37% Y-BOCS				59.8% CGIi
Mallya et al. [38]			1992						McLean Hospital, Harvard Medical School, Belmont, MA, USA						28 with HAM-D<20; 14 placebo vs. 14 fluvoxamine					RCT fluvoxamine vs. placebo ×10 wk															Effect: ↓Y-BOCS from BL;Response: ≥35%↓Y-BOCS from BL										5%↓Y-BOCS									7% Y-BOCS								Fluvoxamine 50→300 mg/day							33% ↓Y-BOCS				43% Y-BOCS
Pigott et al. [39]			1992						NIH Clinical Center, Rockville Pike, Bethesda, MD, USA						17 drug-free; 6 placebo vs. 11 trazodone					DB RCT trazodone vs. placebo ×10 wk															Effect: ↓Y-BOCS scores from BL										10.3% Y-BOCS																	Trazodone 50→ 300 mg/day							12.98%; Y-BOCS
Riddle et al. [21]			1992						Yale Child Study Center, Yale University, New Haven, CT, USA						14 (8.5-16 years); 6 placebo first vs. 7 fluoxetine first					Cross-over randomised study fluoxetine vs. placebo ×20 wk (first 8 wk to one and 12 to the other in random order)															Effect: ↓CY-BOCS scores from BL										26.72% (week 8) CY-BOCS																	Fluoxetine 20 mg/day							44.03% (week 8) CY-BOCS
Stein et al. [40]			1992						Department of Psychiatry, College of Physicians and Surgeons, Columbia University, New York State Psychiatric Institute, New York, NY, USA						35 with ≥56 on SRON and SROC; comorbidity with depression only if OCD primary and dominating; 21 placebo vs. 14 clomi					Multicentre (2 sites) DB RCT to clomi vs. placebo ×10 wk															Effect: Score ↓OCS, SRON, SROC; Response: CGIi 1-2										12% OCS;33.33% SRON;31.63% SROC									19% CGIi								clomi 25→100- 300 mg/day							29.49% OCS;40.49% SRON;29.11% SROC				50% CGIi
Grady et al. [41]			1993						Department of Psychiatry, Duke, University Medical Center; Durham, NC, USA						13 DR (fluoxetine) OCD 80 mg/day ×10 wks; order of administration not specified					DB cross-over to add-on buspirone vs. placebo ×8 wk to unchanged fluoxetine (80 mg/day)															Effect: ↓Y-BOCS score from BL; Response: Y-BOCS ↓≥25% from BL and other unusual criteria										-2.9% (↑Y-BOCS scores)									0% (Y-BOCS)								Buspirone → 60 mg/day added on fluoxetine, 80 mg/die							3.91% (Y-BOCS)				7.7% (Y-BOCS)
Author(s)		Year						Affiliation/ Country						Population (N)												Design							Outcome Measures (Response Criteria)													Placebo Effect (%)													Placebo Responders (%)							Drug(s) or other Treatment than Placebo				Drug Effect (%)				Drug Responders (%)
Hoehn-Saric et al. [42]		1993						Department of Psychiatry and Behavioral Sciences, Johns Hopkins University School of Medicine, Baltimore, MD, USA						21 with NIMH-OC≥9, Y-BOCS≥16; Ham-D≤21; 10 placebo vs. 11 clomi												DB CRT clomi vs. placebo ×10 wk							Effect: ↓NIMH-OC and ↓Y-BOCS scores from BL; Response: not investigated													2.083% NIMH-OC; 5.714% Y-BOCS																				Clomi 25→200 mg/day; min 100 max 300 mg/day				31.521% NIMH-OC; 39.534% Y-BOCS
McDougle et al. [43]		1993						Clinical Neuroscience Research Unit, Yale University School of Medicine, Connecticut Mental Health Center, New Haven, CT, USA						33 DR (failure to reach ↓≥35% drop from BL Y-BOCS scores after fluvoxamine × 8 wk) at flexible doses up to 300 mg/day; 14 placebo vs. 19 add-on buspirone												RCT to add-on buspirone vs. placebo ×8 wk to unchanged fluvoxamine (up to 300 mg/day)							Y-BOCS (response: ↓≥35% from BL); CGIi 1-2; clinician consensus													9.09% (Y-BOCS)													14.29% (Y-BOCS, CGIi, clinician consensus)							Buspirone 15→60 mg/day added on fluvoxamine at the same dose of BL				4.96% (Y-BOCS)				10.52% (Y-BOCS, CGIi, clinician’s consensus)
Montgomery et al. [22]		1993						Imperial College London, Paterson Wing, St Mary's Hospital Medical School, London UK						214; 56 placebo vs. 52 fluoxetine, 20 mg/day vs. 52 fluoxetine, 40 mg/day vs. 54 fluoxetine, 60 mg/day												Multicentre (13 sites) 8-wk DB fluoxetine 20, 40 or 60 mg vs. placebo							Treatment effect: ↓Y-BOCS scores Response: ≥↓25% from BL and CGIi 1-2													17.5% Y-BOCS													26%							Fluoxetine 20,40 and 60 mg/day				21.6%20.5%28.6% Y-BOCS				36%48%47%
McDougle et al. [44]		1994						Department of Psychiatry, Yale University School of Medicine, Yale Child Study Center, New Haven, CT, USA						34 (failure to reach ↓≥35% drop from BL Y-BOCS scores after fluvoxamine × 7 wk) with or without tics; 17 add-on placebo vs. 17 add-on haloperidol												Double-blind DR to fluvoxamine since 7 wk randomised ×4 wk to add-on haloperidol or placebo							Response: Y-BOCS ↓≥35% from BL and final Y-BOCS≤16; CGIi 1-2; and consensus of clinician. Two criteria met: partial responder; all three met: marked responder													7.63%													0%							Fluvoxamine up to 300 mg/day; dose unaltered during trial; add-on haloperidol 2→max10 mg/die (mean, 6.2 mg)				27.17%				65%
Tollefson et al. [23]		1994						Psychopharmacology Division, Eli Lilly & Co., Indianapolis, Ind, USA						355; 89 placebo vs. 87 fluoxetine 20 mg, vs. 89 fluoxetine 40 mg vs. 90 fluoxetine 60 mg												Multicentre (8 sites), DB fluoxetine at fixed doses vs. placebo × 13 wk							Y-BOCS ↓ from initial score for effect;Y-BOCS ↓≥35% for response													-1.2%(mean Y-BOCS scores ↑)													8.5%							Fluoxetine 20 mg/day40 mg/day60 mg day				15.24%20.26%25.99%				32.1%32.4%35.1%
Tollefson et al.* [24]		1994						Psychopharmacology Division, Lilly Research Laboratories, Eli Lilly & Co., Indianapolis, Indiana, USA						76 responders of the previous study with their treatment confirmed; 6 placebo vs. 23 fluoxetine 20 mg, vs. 21 fluoxetine 40 mg vs. 26 fluoxetine 60 mg												Multicentre (8 sites), DB fluoxetine at fixed doses vs. placebo × 6 months (extension)							Y-BOCS ↓ from initial score for effect;Y-BOCS ↓≥35% for response													-6.25%(mean Y-BOCS scores ↑), but at BL patients were not OCD													50% Y-BOCS							Fluoxetine 20 mg/day40 mg/day60 mg day				10.09%11.32%23.91% Y-BOCS, but at BL patients were not OCD				47% Y-BOCS
Author(s)		Year						Affiliation/ Country								Population (N)										Design							Outcome Measures (Response Criteria)													Placebo Effect (%)													Placebo Responders (%)							Drug(s) or other Treatment than Placebo				Drug Effect (%)				Drug Responders (%)
Greist et al. [45]		1995						Dean Foundation for Health, Research, and Education, Madison, Wisconsin, USA								325; 84 placebo vs. 241 sertraline: 80 sertraline 50 mg/day; 81 sertraline 100 mg/day; 80 sertraline 200 mg/day										Multicentre (11 sites), DB parallel comparison of three dosages of sertraline vs. placebo ×12 wk							Effect: ↓Y-BOCS score from BL;Response: Y-BOCS ↓≥25% from BL CGIi 1-2													14.6% ↓Y-BOCS													30% CGIi							Sertraline, pooled50,100, and200 mg/day				23.4%27.37%22.83%33.02% ↓Y-BOCS				38.9% CGIi
Fux et al. [46]		1996						Ministry of Health Mental Health Center, Faculty of Health Sciences, Ben Gunion University of the Negev, Beersheva, Israel								13 partial or complete non-responders to clomi/SSRIs or with severe side effects; 6 placebo first, 7 inositol first										Cross-over RCT inositol vs. placebo ×6 wk; cross-over ×6 wk without washout							Response criteria not specifically stated, but data given so that we extrapolated % ↓Y-BOCS scores for response and calculated % of patients who achieved ↓Y-BOCS≥35% (or 25%) from BL for response													1.55% Y-BOCS; Placebo first at 6 wk: 12.4%													0% ↓Y-BOCS≥ 35%Placebo first at 6 wk: 0% ↓Y-BOCS≥ 25%							Oral inositol 18 g/day				21.104% Y-BOCS; Inositol first at 6 wk: 25.01%				30.77% ↓Y-BOCS≥ 35%; Inositol first at 6 wk: 28.57%(57.14% ↓Y-BOCS≥ 25%)
Goodman et al. [47]		1996						Department of Psychiatry, University of Florida College of Medicine, Gainsville, FL, USA								156; 78 placebo vs. 78 fluvoxamine										Multicentre (4 sites), DB RCT fluvoxamine vs. placebo ×10 wk							Effect: ↓Y-BOCS score from BL; ↓NIMH-OC score from BL;Response: CGIi 1-2													5.42% Y-BOCS;3.89% NIMH-OC													8.6%							Fluvoxamine started at 50 mg/day → max 300 mg/day				21.14% Y-BOCS;19.101% NIMH-OC				43.4%
Nakajima et al. [48]		1996						Department of Neuropsychiatry, Kyoto Perfectural University of Medicine, Kawaramachi-Hirokoji, Kamigyoku, Kyoto, Japan								27 fluvoxamine 300 mg/die vs. 34 fluvoxamine 150 mg/die vs. 33 placebo										DSM-III-R OCD patients assigned to one of three groups: high-dose fluvoxamine vs. low-dose fluvoxamine vs. placebo × 8 wk							Y-BOCS scores (%↓Y-BOCS from BL; response ↓≥35% from BL)													7.25%													36.36%							Fluvoxamine 150 mg/day300 mg/day				28.33%;29.96%				79%;77.8%
Zohar et al. [49]		1996						Chaim Sheba Medical Centre, Tel-Hashomer, Israel, and Sackler Medical School, Tel Aviv University, Israel								399; 99 placebo vs. 201 paroxetine vs. 99 clomi										Multicentre-multinational (? sites), DB paroxetine vs. clomi vs. placebo × 12 wk							Y-BOCS ↓≥25%													18.87%													35.4%							Paroxetine 20-60 mg/day; Clomi 50-250 mg/day				30.77%;32%				55.1%;55.3%
Jenike et al. [50]		1997						Department of Psychiatry, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA								64; 21 placebo vs. 20 phenelzine vs. 23 fluoxetine										10-wk fluoxetine vs. phenelzine vs. placebo							↓Y-BOCS scores from BL as treatment effect													1%																				Phenelzine 60 mg/day,Fluoxetine 80 mg/day				9.4%;14.7%
Author(s)		Year						Affiliation/ Country								Population (N)								Design									Outcome Measures (Response Criteria)													Placebo Effect (%)													Placebo Responders (%)							Drug(s) or other Treatment than Placebo				Drug Effect (%)				Drug Responders (%)
Lindsay et al. [51]		1997						Clinical Research Unit for Anxiety Disorders, School of Psychiatry, University of New South Wales at St Vincent’s Hospital, Sydney, NSW, Australia								18 (13 drug-free, 5 unresponsive to clomi or fluoxetine); 9 placebo (anxiety management) vs. 9 CBT								Parallel assignment to CBT (ERP) vs. anxiety management assumed as the placebo (5 1-h sessions per wk × 3 wk); control group also did homework									Treatment effect: ↓Y-BOCS from BL; ↓MOCI from BL; ↓PADUA scores from BL; Response criteria: not provided													-5.93% (↑Y-BOCS); 12.002% MOCI; 15.201% PADUA																				CBT in 15 sessions divided in three weeks, about 1 hour per session, for both ERP and anxiety management (hyperventilation and respiration and relaxation control with no cognitive restructuring and exercise at home)				61.67% Y-BOCS;43.76% MOCI;50.99% PADUA
Ushijima et al. [52]		1997						Department of Psychiatry, Jikei University School of Medicine, Nishishinbashi, Minato-ku, Tokyo, Japan								33 150 mg/die fluvoxamine vs. 29 300 mg/die fluvoxamine vs. 42 placebo								Multicentre, DB fluoxetine at fixed doses vs. placebo × 8 wk									Y-BOCS ↓ from initial score for effect;Y-BOCS ↓≥35% for response													11.66% (mean Y-BOCS ↓)													38.1% Y-BOCS							Fluvoxamine 150 mg/day300 mg/day				23.9%24.05% (mean Y-BOCS ↓)				55.17%51.51% Y-BOCS
Fallon et al. [53]		1998						Department of Psychiatry, Columbia University, Division of the New York State Psychiatric Institute, New York, NY, USA								54, 29 DR placebo vs. 25 DR i.v. clomi								DR to oral clomi, randomised to i.v. clomi vs. placebo × 14 days									Effect: %↓Y-BOCS from BL; %↓CGIs from BL; %↓NIMH-OC from BL;response ↓≥25% from BL; CGIi 1-2													3.3% Y-BOCS; 0% CGIs; 0% NIMH-OC													0% CGI; 0% Y-BOCS							Intravenous (i.v.) Clomi 250 mg/day				11.8%Y-BOCS; 10.1% CGIs; 9.565% NIMH-OC				20.7%CGIi21.4%Y-BOCS
Li, J. et al. [54]		1998						Mental Health Center of Sichuan province, Mianyang, China								42; 12 placebo vs. 15 clomi vs. 15 paroxetine								DB CCMD-2 OCD paroxetine 20→80 mg/day vs. 50→300 mg/day vs. placebo ×4 wk									Effect: ↓Y-BOCS from BL; unclear response criteria													10.06%																				20-80 mg/day paroxetine,50-300 mg/day clomi				51.32%48.71%
March et al. [55]		1998						Departments of Psychiatry and Psychology, Duke University Medical Center, Durham, NC, USA								189 children and adolescents; 95 placebo vs. 92 sertraline								Multicentre (12 sites), RCT Sertraline flexible doses vs. placebo ×12 wk									Effect: ↓CY-BOCS from BL;Response: ≥25%↓CY-BOCS from BL													15.31% ↓CY-BOCS													37% ≥25%↓CY-BOCS							Sertraline 25-50→max200 mg/day				29.05%↓CY-BOCS				53% ≥25%↓CY-BOCS
Kronig et al. [56]		1999						Department of Psychiatry, Hillside Hospital of LIJMC, Glen Oaks, New York, NY, USA								167; 81 placebo vs. 86 sertraline								Multicentre (10 sites), Double-blind sertraline 50-200 mg/die randomised vs. PLC ×12 wk									CGIi 1-2 (response), ↓Y-BOCS from BL; ↓NIMH from BL (primary outcome)													17.16% Y-BOCS													23.45% CGIi							Sertraline 50-200 mg/day				38.08%				41.8%
Author(s)		Year						Affiliation/ Country								Population (N)								Design									Outcome Measures (Response Criteria)													Placebo Effect (%)													Placebo Responders (%)							Drug(s) or other Treatment than Placebo				Drug Effect (%)				Drug Responders (%)
Dannon et al. [57]		2000						Psychiatry Department, Chaim Sheba Medical Center, Tel Hashomer, Israel								14 DR to 60 mg/day paroxetine ×≥15 wks (Y-BOCS ↓<25%), 6 placebo vs. 8 pindolol								DB RCT of DR, ×6 wk to add-on pindolol or placebo									Unresponsiveness: Y-BOCS ↓<25% from BL;Effect of treatment: ↓Y-BOCS scores from BL													7.69%																				Paroxetine 60 mg/day; add-on pindolol 7.5 mg/die				25.69%
McDougle et al. [58]		2000						Department of Psychiatry, Section of Child and Adolescent Psychiatry, Indiana University School of Medicine, Indianapolis, IN, USA								36 SSRI-refractory; 16 placebo vs. 20 add-on risperidone								Double-blind refractory to clomi, fluvoxamine, sertraline, fluoxetine, and paroxetine (failure to reach ↓≥35% drop from BL Y-BOCS scores), randomised ×6 wk to add-on risperidone or placebo									Response: Y-BOCS ↓≥35% from BL and final Y-BOCS≤16; CGIi 1-2; and consensus of clinician. Two criteria met: partial responder; all three met: marked responder													9.42%													0							Clomi, fluvoxamine, sertraline, fluoxetine, or paroxetine at fixed dose; add-on risperidone 1-6 mg/die (mean, 2.2 mg)				31.8%				50%
Geller et al. [59]		2001						Pediatric OCD Clinic, McLean Hospital, Belmont,MA, USA								103 with ≥4 on the CGIs and ≥16 on the CY-BOCS (7-18 years); 32 placebo vs. 71 fluoxetine								Multicentre (21 sites), 13-wk, DB RCT 2:1. Fluoxetine vs. placebo									CY-BOCS scores (response ↓≥40% from BL)													19.7%													25%							Fluoxetine, mean 24.6 mg/day				38.7%				49%
Montgomery et al. [60]		2001						Imperial College of Science, Technology and Medicine, London, UK								401; 101 placebo vs. 102 20 mg citalopram, vs. 98 40 mg citalopram, vs. 100 60 mg citalopram								Multicentre (53 sites), multinational (12 countries); randomisation to citalopram vs. placebo ×12 wk									Treatment effect: ↓Y-BOCS scores Response: ≥↓25% from BL													22.05%													36.6%							Citalopram 20 mg/day40 mg/day or 60 mg/day				33.47%34.23%40.15%				57.4%; 52%;65%;
Riddle et al. [61]		2001						Division of Child and Adolescent Psychiatry, Johns Hopkins Hospital, Baltimore, MD, USA								120 (8-17 years); 63 placebo vs. 57 fluvoxamine								Multicentre (17 sites) DB RCT fluoxetine vs. placebo × 10 wk									Effect: ↓CY-BOCS scores from BL; Response: CY-BOCS ↓≥25% from BL													13.63% CY-BOCS													26%							Fluvoxamine 25→ 50-max 200 mg/day				24.79% CY-BOCS				42.1%
Romano et al. [62]		2001						Lilly Research Laboratories, Lilly Corporate Center, Indianapolis, IN, USA								71 responders (out of 130) to 20-wk 20, 40 or 60 mg fluoxetine with ↓≥25% drop from BL; 35 placebo vs. 36 fluoxetine								Multicentre (11 sites), randomisation to fluoxetine as before vs. placebo × 52 wk									No relapse													62%																				Fluoxetine 20, 40 or 60 mg/day				82.5%; only 60 mg/die fluoxetine different from placebo
Author(s)		Year						Affiliation/ Country			Population (N)									Design													Outcome Measures (Response Criteria)													Placebo Effect (%)													Placebo Responders (%)							Drug(s) or other Treatment than Placebo				Drug Effect (%)				Drug Responders (%)
Atmaca et al. [63]		2002						Fırat Üniversitesi Hastanesi, Fırat Tıp Merkezi, Psikiyatri Anabilim Dalı, Elazig, Turkey			27 DR (one 3-month trial of 25→300 mg/day clomi or 50→300 mg/day fluvoxamine or 20→80 mg/day fluoxetine yielded CGIi ≥3 and Y-BOCS ≥18 and clinician consensus) to SSRIs or clomi ×2 months), 13 placebo vs. 14 DR to quetiapine add-on									Single-blind randomisation of DR patients with stabilised SSRI or clomi to add-on quetiapine or placebo × 8 wk													Treatment effect: ↓Y-BOCS score from BL; ↓CGIs from BL; response: significant improvement, ≥60% ↓Y-BOCS score from BL; partial improvement, ≥30-59% ↓Y-BOCS score from BL plus clinician consensus													10.08% (↓Y-BOCS);15.88% (↓CGIs)													0%							Stable SSRI or clomi;add-on quetiapine flexible, i.e., 50 mg/day → ↑25 mg/day when Y-BOCS did not ↓ by 2 from previous evaluation				44.4% (↓Y-BOCS);47.161% (↓CGIs)				71.4%
Greist et al. [64]		2002						Healthcare Technology Systems, University of Wisconsin, Madison, WI, USA			218 (age 15-80), Y-BOCS≥16, not comorbid with psychoses and Tourette’s; 75 placebo (relaxation) vs.74 computer-guided vs. 69 therapist-guided									Multicentre (8 sites), parallel comparison of three psychotherapies, computer-assisted CBT, therapist-conducted CBT vs. relaxation assumed as the placebo × 10 wk (therapists not raters)													Effect: ↓Self-rated Y-BOCS score from BL; Response: Y-BOCS ↓≥25% from BL CGIi 1-2; PGIi 1-2													6.6% ↓Self-rated Y-BOCS Relaxation, 1-h/day ×10 weeks													15% PGIi; 14% CGIi							Computer-guided CBT, 9 steps: 1-3 education and assessment; 4-9 self ERP ×1 h or more;Clinician-guided CBT, one 1-hour weekly session ×11 weeks ERP + homework				22.77% 30.16% ↓Self-rated Y-BOCS				38% PGIi;38% CGIi 58% PGIi 60% CGIi
Koran et al. [65]		2002						Department of Psychiatry, Stanford, CA, USA			223 responders to 16 or 52-wk 50-200 mg/day sertraline with Y-BOCS ↓≥25% from BL and CGIi ≤3; 114 placebo vs. 109 sertraline									Multicentre (21 sites), DB randomisation to flexible sertraline or to placebo × 28 wk													No relapse, defined as either ↑Y-BOCS scores ≥5 from randomisation and total score of ≥20 and ↑CGIi score of ≥1; Insufficient clinical response													76%																				Sertraline flexible doses 50-200 mg/day				91%
Liebowitz et al. [66]		2002						New York State Psychiatric Institute and Department of Psychiatryat Columbia University, New York, NY, USA			43 (children 6-18 years); 22 placebo vs. 21 fluoxetine (acute); 7 placebo vs. 11 fluoxetine (extension)									Multicentre (2 sites), RCT fluoxetine vs. placebo ×8 wk (acute phase) → responders extension ×8 wk													CGIi 1 or 2 (response), ↓CY-BOCS from BL (primary outcome)													22.12% acute;54.6% extension													31.81% (CGIi)							Fluoxetine 60-80 mg/die				34.62% acute; 74.36% extension				57.14% CGIi
Author(s)	Year						Affiliation/ Country								Population (N)						Design									Outcome Measures (Response Criteria)													Placebo Effect (%)									Placebo Responders (%)									Drug(s) or other Treatment than Placebo									Drug Effect (%)				Drug Responders (%)
Geller et al. [67]	2003						Obsessive Compulsive Disorder Program and Pediatric Psychopharmacology Research Program, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA								8-17 yr-old children who responded to 16-wk 10-60 mg/day paroxetine; 98 placebo vs. 95 paroxetine						Multicentre (? sites); paroxetine-responder children aged 8-17 randomised to paroxetine as before vs. placebo ×16 wk									CY-BOCS (response ↓≥25% from BL and CGIi 1 or 2);No relapse; relapse defined as ↑CGI by 1 in two follow-up visits or ↑CGI by 2 at any time or CGI≥5 at any time																						56.1%									Paroxetine flexible doses (10-60 mg/die)									65.3%
Hollander et al. [68]	2003						Department of Psychiatry, The Mount Sinai School of Medicine, New York, NY, USA								27, 10 placebo vs. 17 clonz						Multicentre (2 sites), DR + drug- naïve patients randomised 2:1 to clonz or placebo × 10 wk									CGIi 1-2													2.6%Y-BOCS									22% CGIi									Clonz 3-6 mg/day									7% Y-BOCS				6.2% CGIi
Hollander et al. [69]	2003						Department of Psychiatry, The Mount Sinai School of Medicine, New York, NY, USA								16 DR (CGIi≥3 after at least two trials of SSRIs, clomi or venlafaxine for 12 wks), 6 placebo vs. 10 add-on risperidone						DB RCT of DR patients to risperidone vs. placebo × 8 wk added on stable previous SSRI, venlafaxine or clomi									Effect: ↓Y-BOCS score from BLResponse: Y-BOCS ↓≥25% and CGIs ↓≥2 points from BL													4.53% Y-BOCS									0% Y-BOCS and CGIi									Risperidone 0.5→max 3 mg/day add-on to minimum mg/day: 200 clomi, 60 fluoxetine, 150 fluvoxamine, 150 sertraline, 60 citalopram or 325 venlafaxine									20.89% Y-BOCS				40% Y-BOCS and CGIi
Hollander et al. [70]	2003						Department of Psychiatry, The Mount Sinai School of Medicine, New York, NY, USA								89 placebo vs. 88 paroxetine, 20 mg/day, vs. 86 paroxetine, 40 mg/day, vs. 85 paroxetine, 60 mg/day						Multicentre (15 sites), paroxetine, three fixed doses with an about 1:1:1:1 randomisation DB vs. placebo × 12 wk									Effect: ↓Y-BOCS score from BLResponse: Y-BOCS ↓≥25% and CGIs ↓≥2 points from BL													13% Y-BOCS									Not provided									Paroxetine 20,40,60 mg/day									16%25%29% Y-BOCS				Not provided
Hollander et al. [70]	2003						Department of Psychiatry, The Mount Sinai School of Medicine, New York, NY, USA								105 responders to paroxetine randomised to placebo (N=52) or to paroxetine (N=53) 20-60 mg/day						Multicentre (15 sites), Paroxetine responders to flexible paroxetine doses (20-60 mg/day) or placebo × 6 months									Non-relapse: relapse defined as return of Y-BOCS to BL values or ≥1↑CGI at any time-point																						41.2% Y-BOCS/ CGIi									Paroxetine 20, 40 or 60 mg/day													62.3% Y-BOCS/ CGIi
Hollander et al. [71]	2003						Department of Psychiatry, The Mount Sinai School of Medicine, New York, NY, USA								253; 126 placebo vs. 127 fluvoxamine controlled-release						Multicentre (5? sites) RCT Fluvoxamine controlled-release 100-300. placebo × 12 wk									Effect: ↓Y-BOCS score from BLResponse: CGIi 1 or 2													19.01% Y-BOCS									23% CGIi									Fluvoxamine 100 → 100-300 mg/day									34.56% Y-BOCS				44% CGIi
Author(s)	Year						Affiliation/ Country										Population (N)										Design														Outcome Measures (Response Criteria)										Placebo Effect (%)								Placebo Responders (%)								Drug(s) or other Treatment than Placebo			Drug Effect (%)				Drug Responders (%)
Bystritsky et al. [72]	2004						Department of Psychiatry and Biobehavioral Sciences, Anxiety Disorders Program, University of California at Los Angeles (UCLA) School of Medicine, Los Angeles, CA, USA										26 DR (unchanged after two adequate antidepressant trials and a course of behavioural psychotherapy), 13 placebo vs. 13 risperidone										DR to clomi or SSRIs with dose unchanged randomised × 6 wk to add-on olanzapine or placebo														Effect: % ↓Y-BOCS; Response: Y-BOCS ↓≥25% from BL										-1.99% (↑Y-BOCS from BL)								0% (↓≥25% Y-BOCS)								Full-dose clomi, fluoxetine, sertraline, paroxetine;add-on olanzapine 2.5 → 5-20 mg/day			17.36% (↓Y-BOCS from BL)				46% (↓≥25% Y-BOCS)
Denys et al. [73]	2004						Rudolf Magnus Institute of Neuroscience (DD), Department of Psychiatry, University Medical Center Utrecht, Utrecht, The Netherlands										20 DR to SSRIs to add-on placebo; 20 to add-on quetiapine										DR to SSRIs (failure to achieve ↓≥25% of Y-BOCS scores from BL) randomised to add-on quetiapine vs. placebo × 8 wk														Y-BOCS (response ↓≥35% drop from BL); CGIi 1-2										6.8% Y-BOCS; 7,5% CGIi								10% CGIi								Various SSRIs at various doses plus quietapine (B) 200-300 mg/day			31.9% Y-BOCS27,5% CGIi				40% CGIi
Fux et al. [74]	2004						BeerSheva Mental Health Center, Ben Gurion University of the Negev, Beer-Sheva, Israel										11 with unsatisfactory response to SSRI during last 2 months: 5 placebo add-on vs. 6 EPA add-on										Cross-over RCT, EPA vs. placebo added-on highest tolerated SSRI dose														Response criteria not specified; assessment performed with Y-BOCS										32.3%																Various SSRIs at various doses; EPA 2g/day			28.8%
Geller et al. [75]	2004						Pediatric OCD Clinic, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA										203 (7-17 years); 105 placebo vs. 98 paroxetine										Multicentre (34 sites); paroxetine vs. placebo × 10 wk														Effect: ↓CY-BOCS score from BL;Response: ↓≥25% CY-BOCS from BL										21.1%								41.2%								Paroxetine 10-50 mg/die			36%				64.9%
Kamijima et al. [76]	2004						Showa University School of Medicine, Shinagawa-ku, Tokyo, Japan										188; 94 placebo vs. 94 paroxetine										Multicentre (? sites), DB paroxetine vs. placebo × 12 wk														Effect: ↓Y-BOCS scores from BL; response: ↓≥25% Y-BOCS from BL; CGIi 1-2										14.8% Y-BOCS								23.7% CGIi								Paroxetine 20-50 mg/day			33.4% Y-BOCS				50% CGIi
Shapira et al. [77]	2004						Department of Psychiatry, University of Florida, College of Medicine, Gainesville, FL, USA										44 partial/non responders to fluoxetine, 22 placebo vs. 22 add-on olanzapine										DR (partial/non-responders, i.e., <↓25% Y-BOCS score from BL or <16 total Y-BOCS or symptomatic) to 8-wk double-blind fluoxetine randomised × 6 wk to add-on olanzapine or placebo														Treatment effect: % ↓Y-BOCS scores from BL; Response: Y-BOCS ↓≥25% from BL										3.8%								41% ↓≥25% from BL; (18% ↓≥35% Y-BOCS from BL)								Fluoxetine 40 mg/day; Olanzapine 5-10 mg/day			5.1%				41% ↓≥25% from BL; (23% ↓≥35% Y-BOCS from BL)
Author(s)	Year						Affiliation/ Country								Population (N)												Design													Outcome Measures (Response Criteria)							Placebo Effect (%)									Placebo Responders (%)								Drug(s) or other Treatment than Placebo						Drug Effect (%)				Drug Responders (%)
The Pediatric OCD Treatment Study (POTS) Team [78]	2004						Department of Psychiatry, Duke University Medical College, Durham, NC, USA								112 children and adolescents (7-17 yrs.) CY-BOCS≥16; 28 placebo vs. 28 sertraline vs. 28 CBT vs. 28 sertraline + CBT placebo												Multicentre (3 sites), balanced, masked RCT (biased by patient preference) of sertraline vs. CBT vs. sertraline + CBT vs. placebo × 12 wk													Effect: ↓CY-BOCS from BL; Response: CY-BOCS≥10							14.68%									3.6%								Sertraline 25→max200 mg/day;14 CBT sessions with psychoeducation, cognitive training mapping OCD target symptoms and ERP;Sertraline + CBT						29.79% 46.15% 52.84%				21.4% 39.3% 53.6%
Carey et al. [79]	2005						MRC Research Unit on Anxiety Disorders, University of Stellenbosch, Cape Town, South Africa								42 SSRI-resistant patients defined as not responding to at least two adequate trials (12 wks); 21 SSRI+placebo, 20 SSRI+ quetiapine												Multicentre RCT (5 sites); DR (CGI≥3 or ↓Y-BOCS ≤25% from before treatment) to SSRI randomised to add-on quetiapine vs. placebo × 6 wk													Treatment effect: % ↓Y-BOCS score from BLResponse: ↓Y-BOCS ≥25% from BLand CGIi 1-2							26%Y-BOCS22.6%CGIi									47.6%								Various SSRIs at various doses;Add-on quetiapine,≈200 mg/day						26.9%Y-BOCS21,1%CGIi				40%
Erzegovesi et al. [80]	2005						Department of Neurosciences, San Raffaele Hospital, Vita-Salute San Raffaele University, Milan, Italy								39 patients stabilised on fluvoxamine 150-300 mg/day; 19 placebo vs. 20 add-on risperidone; 20 DR (10 placebo; 10 risperidone) and 19 responders (9 placebo; 10 risperidone)												DR to fluvoxamine (failure to reach ↓≥35% drop from BL Y-BOCS score) and responders to open fluvoxamine randomised × 6 wk to add-on risperidone or placebo													Y-BOCS scores (response ↓≥35% drop from BL);(Responders and non-responders here refer to open fluvoxamine and not to the add-on)							13.89%;Responders 27.63%; Non-responders: 6.82% Y-BOCS									Non-resp 20% Y-BOCS								Fluvoxamine 150-300 mg/die; risperidone 0.5 mg/die						18.24%;Responders, 3.82%; non-responders, 25.57%				Non-responders, 50%
Fineberg et al. [81]	2005						Department of Psychiatry, Queen Elizabeth II Hospital, Hertfordshire, Welwyn Garden City, Department of Psychology, University of Hertfordshire, Hatfield, UK								21 DR (Y-BOCS ≥18 and ↓≥25% Y-BOCS after ≥12 wks SSRI at max tolerated dose); 10 placebo, 11 quetiapine add-on												Add-on of quetiapine vs. placebo × 16 wk on stable SSRI at max tolerated dose													Effect: %↓Y-BOCS from BL;response ↓≥25% from BL							6%									10%								Quetiapine 25→max 400 mg/die + SSRI (citalopram, sertraline, paroxetine)						14%				27.27%
Foa et al. [82]	2005						Center for the Treatment and Study of Anxiety, University of Pennsylvania, Philadelphia, PA, USA								122, 26 placebo vs. 36 clomi, vs. 29 ERP, vs. 31 clomi + ERP												Multicentre (3 centres), 12-wk RCT; Clomi vs. placebo vs. ERP vs. clomi + ERP													Effect: ↓Y-BOCS from BLResponse: CGIi <3							11.2% Y-BOCS6% CGIi									8%								Clomi 200 mg/dayERP: 2h exposure sessions, 5 times a week + daily exposure and ritual prevention homework max 2hCombined Clomi + ERP						30.7% Y-BOCS; 19.6% CGIi55.28% Y-BOCS; 43.75% CGI58.66% Y-BOCS; 40.82 CGIi				41.67% 62.07% 67.74%
Author(s)	Year						Affiliation/ Country								Population (N)								Design																Outcome Measures (Response Criteria)											Placebo Effect (%)					Placebo Responders (%)									Drug(s) or other Treatment than Placebo								Drug Effect (%)	Drug Responders (%)
Kobak et al. [83]	2005						Dean Foundation For Health, Research and Education, Middleton, WI, USA								60 with Ham-D <16; 30 placebo vs. 30 St. John’s wort (Hypericum perforatum)								Multicentre (4 sites), DB CRT of flexible Hypericum or placebo × 12 wk																Effect: ↓Y-BOCS scores from BL; response: CGIi 1-2											20.665% Y-BOCS					16.7% CGIi									Hypericum perforatum (Saint John’s wort) flexible doses 600→1800 mg/day								23.306% Y-BOCS	17.9% CGIi
Koran et al. [84]	2005						Department of Psychiatry and Behavioral Sciences, Stanford University Medical Center, Stanford, CA, USA								23 DR with OCD since at least 3 years, with Y-BOCS ≥20 and poor response to at least two SSRI trials (≥8 wks at full dose); 32 placebo								Add-on to stable antidepressant therapy of morphine, lorazepam or placebo × 7 wk: 2 wk of each in random order; cross-over, but only first part considered (prior to cross-over)																Effect: ↓Y-BOCS scoreResponse: Y-BOCS (response ↓≥25% from BL)											7%					0%									Morphine 30 mg/week 1 →15-45 mg/week adjustmentLorazepam 1 mg →0.5-2 mg/week								27% 6%	30.43% 17.4%
Li, X. et al. [85]	2005						Department of Psychiatry and Behavioral Neurobiology, University of Alabama at Birmingham, Birmingham, AL, USA								13 DR to ≥12-wk SSRIs (symptomatic; ≥10 score on Y-BOCS items 1-5; ≥16 total Y-BOCS score); 5 placebo first, 5 haloperidol first, 3 risperidone first								DB cross-over of DR on stable antidepressant, randomised × 9 wk to add-on risperidone, haloperidol or placebo (1 wk placebo→2 wk placebo or haloperidol or risperidone→1 wk placebo→2 wk cross-over→ 1wk placebo→ 2 wk cross-over)																Effect: ↓Y-BOCS from BL; no criteria for response											23.96%														≥40 mg fluoxetine, ≥200 mg fluvoxamine, ≥100 mg sertraline; dose unaltered during trial;add-on haloperidol 2 mg/dayrisperidone 1 mg/day								48.62% 36.29%
Nakatani et al. [86]	2005						Department of Neuropsychiatry, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan								28 DR; 8 placebo + autogenic training vs. 10 behaviour therapy + placebo vs. 10 fluvoxamine + autogenic training								DR patients assigned to one of three groups: behavioural therapy + placebo; fluvoxamine + autogenic training; autogenic training + placebo ×12 wk																Y-BOCS scores (response ↓≥35% from BL), CGIi 1-2; responsiveness to placebo was taken as the response of Y-BOCS scores and the CGIi criterion of the placebo + autogenic training group											6.88%					0%									Fluvoxamine 150-200 mg/day								28.87%	30%
Buchsbaum et al. [87]	2006						Mount Sinai School of Medicine, New York, NY, USA								16 DR, 6 placebo vs. 10 risperidone								DR to clomi, fluvoxamine, sertraline, fluoxetine, paroxetine, citalopram, and venlafaxine randomised double-blind ×8 wk to add-on risperidone or placebo																Response: Y-BOCS ↓≥25% from BL; CGIi 1-2											4.53%					0% (CGIi) 0% (Y-BOCS)									Actual minimum daily doses: 200 mg clomi, 60 fluoxetine, 150 fluvoxamine, 150 sertraline, 60 citalopram, 325 venlafaxine;add-on risperidone 0.5 → 3 mg/day								20.89%	40% (CGIi) 40% (Y-BOCS)
Author(s)	Year					Affiliation/ Country				Population (N)										Design																	Outcome Measures (Response Criteria)													Placebo Effect (%)					Placebo Responders (%)									Drug(s) or other Treatment than Placebo								Drug Effect (%)	Drug Responders (%)
O’Connor et al. [88]	2006					Fernand-Seguin Research Centre, Louis-H. Lafontaine Hospital, Montreal, QC, Canada				21 with Y-BOCS >16; 10 placebo vs. 11 fluvoxamine (first protocol); thereafter, 43 (comprising the first 21) received CBT vs. 10 CBT alone, vs. 12 CBT + stabilised on antidepressant										DB, random allocation to fluvoxamine vs. placebo ×5 months (first protocol); after conclusion or stabilisation (on or off-drug) of another population, all receive CBT; CBT to drug naïve patients, CBT + stabilised on drug, CBT to previous placebo group, CBT to previous fluvoxamine group × 5 months (20 sessions) (second protocol)																	Treatment effect: ↓Y-BOCS scores Response: >↓35% Y-BOCS from BL													6.96% Y-BOCS					0%									Fluvoxamine 100 → max 300 mg/day;CBT – ERP 1 weekly session for a total of 20 over 5 months;CBT + fluvoxamine								15.19% 53.33%(not used in analyses)42.69% Y-BOCS (not used in analyses)	9%
Fineberg et al. [89]	2007					Postgraduate Medical School, University of Hertfordshire, Hatfield, UK				320 responders to 16-wk 10 or 20 mg escitalopram with Y-BOCS ↓≥25% from BL; 157 placebo vs. 163 escitalopram 10 or 20 mg/day										Multicentre (62 sites), multi-national (14 countries), DB RCT escitalopram 10 vs. 20 vs. placebo × 24 wk																	No relapse, defined as either an increase in Y-BOCS scores ≥5 from randomisation or as unsatisfactory treatment effect (lack of efficacy) judged by investigator;Responders: Y-BOCS ↓≥25% from BL													48% no relapse					72% Y-BOCS									Escitalopram fixed dose 10mg/die or 20mg/die								77% no relapse	90% Y-BOCS
Stein et al. [90]	2007					University of Cape Town, Department of Psychiatry, Groote Schuur Hospital, Cape Town, South Africa				458; 114 placebo vs. 113 escitalopram 10, 114 escitalopram 20, 117 paroxetine										Double-blind randomized fixed-dose escitalopram × 24 wk to paroxetine or placebo																	Primary outcome: ↓Y-BOCS from BL at week 12; secondary: mean Y-BOCS change from BL at week 24;Remission: Y-BOCS ≤10Response: CGIi 1 or 2; Y-BOCS ↓≥25% from BL													wk12: 30.54% wk24: 38.51% Y-BOCS					wk12 38.5%; wk24 38% (CGIi); wk12 52%; wk24 50% (Y-BOCS)									Escitalopram 10 mg/die or Escitalopram 20 mg/die or Paroxetine 40 mg/die								wk12: 42.97%wk24: 51.77% wk12: 45.64wk24: 51.84% wk12: 42.75wk24: 54.62%	wk12 50% wk24 58% (CGIi)wk12 66% wk24 63% (Y-BOCS)wk12 56% wk24 58.5% (CGIi)wk12 70.5% wk24 70.5% (Y-BOCS)wk12 54% wk24 58% (CGIi)wk12 65% wk24 67% (CGIi)
Amiaz et al. [91]	2008					Division of Psychiatry, Chaim Sheba Medical Center, Tel-Hashomer, Israel				10 DR to SSRIs or clomi ×2 months; 5 placebo first, 5 naltrexone first										DB cross-over of DR patients to ≥2-month clomi or 2 SSRIs × 5 wk to add-on naltrexone or placebo, × 1 wk to add-on placebo only × 5 wk to cross-over add-on naltrexone or placebo																	Treatment effect: ↓Y-BOCS score from BL; ↓CGIs from BL; no response criteria provided													5.46% (↓Y-BOCS);18.42% (↓CGIs)														Stable SSRI or clomi;add-on naltrexone 50 → 100 mg/die								-16.07% (↑Y-BOCS scores);-10.7% (↑CGIs scores)
Author(s)	Year				Affiliation/ Country						Population (N)										Design															Outcome Measures (Response Criteria)														Placebo Effect (%)					Placebo Responders (%)									Drug(s) or other Treatment than Placebo								Drug Effect (%)		Drug Responders (%)
Kordon et al. [92]	2008				Department of Psychiatry and Psychotherapy, University of Luebeck, Luebeck, Germany						40 DR; 20 placebo vs.20 add-on quetiapine										Multicentre (2 sites), DR to SSRI or clomi × 12 wk (failure to reach ↓≥25% drop from BL Y-BOCS scores) randomised to add-on quetiapine or placebo															Y-BOCS (response ↓≥35% drop from BL);CGIi 1-2														15.1% Y-BOCS					30% CGIi									SSRI or clomi at fixed dose; Quietiapine 400-600 mg/day								21.6% Y-BOCS		22% CGIi
Greenberg et al. [93]	2009				New York University School of Medicine, Department of Psychiatry, New York, NY, USA						13 DR (drugs or psychotherapy), with treatment stabilised ×≥12wks, 9 placebo vs. 5 glycine add-on										DB RCT 1:1 of DR to add-on glycine vs. placebo × 12 wk to unchanged regimen (drugs and/or psychotherapy)															Effect: ↓Y-BOCS score from BL; ↓NIMH-OC score from BL;↓CGIs score from BL;Response: Y-BOCS ↓≥35% from BL and CGIi 1-2														4.04% (Y-BOCS);2.44% NIMH-OC;2.36% (CGIs)					0% (Y-BOCS plus CGIi)									Glycine powder dissolved in water plus flavour enhancer → 60 g/day added on stabilised treatment (drug and/or psychotherapy)								24.59% (Y-BOCS);22.22% NIMH-OC;13.04% (CGIs)		40% (Y-BOCS plus CGI)
Sayyah et al. [94]	2009				Department of Psychiatry, Joondi Shapoor University of Medical Sciences, Ahwaz, Iran						44 with Y-BOCS≥21; 20 placebo vs. 24 aqueous extract of Echium amoenum										DB, RCT of 500 mg aqueous extract of Echium amœnum × 6 wk															Effect: ↓Y-BOCS scores from BLResponse: not provided														11.31% Y-BOCS														125 mg aqueous extract of Echium amoenum capsules: 1 morning; 1 afternoon, 2 night								25.55% Y-BOCS
Mowla et al. [95]	2010				Department of Psychiatry, Bushehr University of Medical Sciences, Bushehr, Iran						41 Y-BOCS≥18 to:-N=20 Topiramate;or-N=21 placebo; × 12 wks										12-wk, double-blind, placebo-controlled, randomized trial of 200 mg/day topiramate vs. placebo															Treatment effects: ↓Y-BOCS from BL; response: Y-BOCS ↓≥25% from BL (after 12 weeks)														2.4% (↓Y-BOCS)					0% (Y-BOCS ↓≥25% from BL)									Topiramate (initially 25 mg/day, increased in 25-mg increments weekly to a target dose of 200 mg/day								32% (↓Y-BOCS);		60% (Y-BOCS ↓≥25% from BL)
Storch et al. [96]	2010				Department of Pediatrics, Rothman Center for Neuropsychiatry, University of South Florida, St. Petersburg, FL, USA						30 children and adolescents with OCD (Range 8-17 years); 15 placebo, 15 D-cycloSer										DB RCT of CBT + D-cycloSer vs. CBT + Placebo × 8 wk (10 sessions). 1:1 randomisation															Effect: % ↓ CGI-S, CYBOCS, and ADIS-CSR from BL.no criterion for response														41% (↓CGI-S), 58% (↓ CY-BOCS), 53% (↓ADIS-CSR)														25 or 50 mg of D-cycloSer (depending on patient weight) 1 h before psychotherapy, sessions 4-10								57%(↓CGI-S)72% (↓ CYBOCS)71% (↓ADIS-CSR)
Sayyah et al. [97]	2011				Education Development Center (EDC), Jundishapur University of Medical Sciences, Ahwaz, Iran						52 drug naïve OCD patients randomised to celecoxib (N=27) or placebo (N=25)										DB RCT of fluoxetine 20 mg/day + celecoxib 400 mg/day vs. fluoxetine 20 mg/day + placebo × 8 wk															Effect: % ↓Y-BOCS score from BL; no criterion for response														46.7% (↓Y-BOCS)														Fluoxetine 20 mg/day + Celecoxib 400 mg/day								66.2% (↓Y-BOCS)
Author(s)	Year				Affiliation/ Country														Population (N)										Design									Outcome Measures (Response Criteria)											Placebo Effect (%)									Placebo Responders (%)								Drug(s) or other Treatment than Placebo					Drug Effect (%)		Drug Responders (%)
Muscatello et al. [98]	2011				Section of Psychiatry, Department of Neurosciences, Psychiatric and Anaesthesiological Sciences, Section of Pharmacology, Department of Clinical and Experimental Medicine and Pharmacology, IRCCS Centro Neurolesi “Bonino-Pulejo” University of Messina, Messina, Italy														30 Y-BOCS≥18 toN=16 aripripazole (15 mg/day)N=14 Placebo										16-wk, open-label. flexible-dose (up to 30 mg/day), pilot trial									Treatment effects: ↓Y-BOCS score from BL; partial response (pr): Y-BOCS ↓≥25% from BL; complete response (cr): Y-BOCS ↓≥35% from BL; remission (r) (Y-BOCS ≤16 after 16 weeks)											-2.54% (↓Y-BOCS)									0% (pr); 0%(cr)								Aripiprazole 15 mg/day added to SSRI					28.5% (↓Y-BOCS);		43.7% (pr); 25% (cr)
Berlin et al. [99]	2011				Department of Psychiatry, The Mount Sinai School of Medicine, New York, NY, USA														36 OCD patients with Y-BOCS ≥ 18; 18 randomised to placebo and 18 to topiramate										DB RCT of add-on topiramate (up to 400 mg/d) over continuing SSRI vs. placebo plus SSRIs × 12 wk									Effect: % ↓Y-BOCS score from BL after 12 weeks. No response criteria											16% (↓Y-BOCS from BL to 12 weeks)																	Add-on topiramate titrated over 8 weeks up to 400 mg/day or maximum tolerated dose					38% (↓Y-BOCS)
Pakseresht et al. [100]	2011				Jundishapur University of Medical Sciences, Ahwaz, Iran														31 (18-60 y) to-N=15 extract of Valeriana Officinalis L. (765 mg/day)or-N=16 placebo (30 mg/day) × 8 wks(Y-BOCS≥21)										8-wk double-blind, parallel-group, randomised trial									Treatment effect: ↓Y-BOCS from BL after 8 weeks; no response criteria provided											23.3% (↓Y-BOCS);																	Valeriana Root (Valeriana Officinalis L.) 750 mg/day in three divided doses					43.3% (↓Y-BOCS)
Sayyah et al. [101]	2012				Jundishapur University of Medical Sciences, Ahwaz, Iran														23 drug naïve OCD patients; add-on to fluoxetine, 12 ZnSO4, 11 placebo										DB RCT of 20 mg/day fluoxetine + 440 mg/day ZnSO4 vs. 20 mg/day fluoxetine + placebo × 8 wk									Effect: % ↓Y-BOCS score from BL.no criterion for response											46.84% (↓Y-BOCS)																	20 mg/day Fluoxetine + 440 ZnSO4 mg/day					54.54% (↓Y-BOCS)
Sayyah et al. [102]	2012				Imam General Hospital, Jundishapur University of Medical Sciences, Ahwaz, Iran														32 adult outpatients:N=15 10 mg/day aripiprazoleN=17 placebo										12-wk, double-blind RCT									Treatment effects: ↓Y-BOCS score from BL; response: Y-BOCS ↓≥25% after 12 weeks											17.6% (↓Y-BOCS);									8.3% (Y-BOCS ↓≥25% after 12 weeks)								Aripiprazole 10 mg/day					29.5% (↓Y-BOCS);		53% (Y-BOCS ↓≥25% after 12 weeks)
Afshar et al. [103]	2012				Nour Hoospital, Psychosomatic Research Center, Department of Psychiatry, School of Medicine, Isfahan University of Medical Sciences, Isfahan, Iran														48 DR OCD patients (SSRI/clomi non-responders); 19 NAC, 20 placebo										DB RCT of add-on →2400 mg/day NAC vs. placebo ×12 wk									Effect: % ↓Y-BOCS score from BL. % ↓ CGI-S from BLPartial response: Y-BOCS ↓≥25% from BL;Response: Y-BOCS ↓≥35% from BL											20.7% (↓Y-BOCS)10.4% (↓CGIs)									Response: 15%; No data on partial response								Initial dosage of 600 mg/d of NAC, which doubled weekly to a maximum dose of 2400 mg/d					39.2% (↓Y-BOCS); 24.9% (↓CGIs)		Response: 52.6%No data on partial response given
Author(s)	Year				Affiliation/ Country														Population (N)										Design									Outcome Measures (Response Criteria)											Placebo Effect (%)									Placebo Responders (%)								Drug(s) or other Treatment than Placebo					Drug Effect (%)		Drug Responders (%)
Bruno et al. [104]	2012				Section of Psychiatry, Department of Neurosciences, Psychiatric and Anaesthesiological Sciences, University of Messina, Messina, Italy														33 DR OCD patients (persistent obsessive-compulsive symptoms: despite adequate SSRI trial(s) → Y-BOCS ≥16); 17 lamotrigine, 16 placebo										DB RCT of add-on lamotrigine 100 mg/day vs placebo for 16 wk									Effect: % ↓Y-BOCS score from BL;Partial response: Y-BOCS ↓≥25% from BL;Response: Y-BOCS ↓≥35% from BL											-1.2% (↓Y-BOCS)									Response: 0%								Add-on lamotrigine increased from 25 mg/day to100 mg/day at week 4, in increments of 25 mg/week Maximum dose of 100 mg maintained until the end of the trial					33.8% (↓Y-BOCS)		Partial response: 50%Complete response: 35%
Askari et al. [105]	2012				Psychiatric Research Center, Roozbeh Hospital, Tehran University of Medical Sciences, Tehran, Iran														39 drug naïve OCD patients (Y-BOCS ≥21) randomised to placebo (N=20) or granisetron (N=19)										Multicentre DB RCT of fluvoxamine 100-200 mg/day + granisetron 2 mg/day vs. fluvoxamine 100 - 200 mg/day + placebo × 8 wk									Effect: % ↓Y-BOCS score from BLPartial response: Y-BOCS ↓≥25% from BLResponse: Y-BOCS ↓≥35% from BLRemission: Y-BOCS ≤ 16											34.7% (↓Y-BOCS)									Partial response: 35%; Complete response: 35%; Remission: 35%								Fluvoxamine 100 mg/day × first 4 weeks, 200 mg/day × next 4 weeks + granisetron 2 mg/day					59.1% (↓Y-BOCS)		Partial response: 100%Complete response: 100%Remission: 90%
Ghaleiha et al. [106]	2013				Research Center for Behavioral Disorders and Substance Abuse, Hamadan University of Medical Sciences, Hamadan, Iran														38 patients with diagnosis of OCD and Y-BOCS score ≥21 randomised to add-on placebo (N=19) or memantine (N=19)										DB RCT of fluvoxamine + memantine vs. fluvoxamine + placebo × 8 wk									Effect: % ↓Y-BOCS score from BLPartial response: Y-BOCS ↓≥25% from BLResponse: Y-BOCS ↓≥35% from BLRemission: Y-BOCS ≤16											36.9% (↓Y-BOCS)									Partial or completeresponse: 32%Remission: 32%								Memantine 10 mg/day forthe first week of the trial, then 20 mg/day					57,9% (↓Y-BOCS)		Partial or completeresponse: 100%Remission: 89%
Storch et al. [107]	2013				Department of Pediatrics, University of South Florida,St. Petersburg, FL, USA														47 children and adolescents with OCD (Range 7-17 years) randomised to RegSert (N=14), SloSert (N=17) or placebo (N=16) + CBT for all										DB RCT of sertraline at standard dosing + CBT or sertraline titrated slowly + CBT or placebo + CBT × 18 wk. Patients randomized in a 1:1:1 fashion.									Effect: % ↓CY-BOCS score from BL.Response: CY-BOCS ↓≥30% from BLRemission: CY-BOCS score <10											37.9% (↓Y-BOCS)									Response: 62.5% (CY-BOCS ↓≥30% from BL)Remission: 18.8% (CY-BOCS score below 10)								RegSert: upward titration from 25 mg/day to 200 mg/day in 5 wk.SloSert: upward titration from 25 mg/day to 200 mg/day in 9 wk					RegSert: 34.7% (↓Y-BOCS)SloSert: 35.5% (↓Y-BOCS)		Response: 57.1% for RegSert + CBT; 64.7% for SloSert + CBT; Remission: 42.9% for RegSert + CBT; 23.5% for SloSert + CBT
Haghighi et al. [108]	2013				Research Center for Behavioral Disorders and Substances Abuse, Hamadan University of Medical Sciences, Hamadan, Iran														29 inpatients with diagnosis of OCD and Y-BOCS score ≥21 despite treatment with SSRI or clomi to: memantine (N=14) or placebo (N=15)										DB RCT of add-on memantine 5–10 mg/day vs. placebo × 12 wk									Effect: % ↓Y-BOCS score from BL. % ↓ CGI-S from 4th week; Partial response: Y-BOCS ↓≥25% from BL; Response: Y-BOCS ↓≥35% from BL											15.8% (↓Y-BOCS)13.4% (↓CGIs)									Partial or complete response: 26.6%								Add-on memantine 5–10 mg/day					32.2% (↓Y-BOCS) 30.2% (↓CGIs)		Partial or complete response: 92.8%
Author(s)	Year			Affiliation/ Country														Population (N)										Design									Outcome Measures (Response Criteria)											Placebo Effect (%)												Placebo Responders (%)					Drug(s) or other Treatment than Placebo					Drug Effect (%)			Drug Responders (%)
Storch et al. [109]	2013			Department of Pediatrics, Rothman Center for Neuropsychiatry, University of South Florida,St. Petersburg, FL, USA														34 DR OCD patients (Y-BOCS≥ 19 despite at least 2 adequate SSRI trials) randomised to placebo (N=17) or paliperidone (N=17)										DB RCT of add-on Paliperidone (up to 9 mg/d) vs. placebo × 8 wk									Effect: % ↓Y-BOCS score from BL. %↓CGI-S from BLResponse: Y-BOCS ↓≥35% from BLCGI-i = 1 - 2											15.7% (↓Y-BOCS)18.7% (↓CGI-S)												Response: 29% (Y-BOCS ↓≥35%) 18% (CGIi =1/2)					Add-on Paliperidone starting from 3 mg/day and titrated up to 9 mg/day by week 6 unless not tolerated					29.4% (↓Y-BOCS) 20.1% (↓CGI-S)			Response: 35% (Y-BOCS ↓≥35%)35% (CGIi=1/2)
Rodriguez et al. [110]	2013			New York State Psychiatric Institute, New York, NY, USA; Department of Psychiatry, Columbia University, College of Physicians and Surgeons, New York, NY, USA														15 drug free OCD patients with Y-BOCS ≥ 16 who had failed at least one prior SSRI trial and/or CBT: 8 to ketamine and 7 to placebo										DB Crossover RCT of iv ketamine (0.5 mg/kg) vs. iv saline spaced at least 1-wk apart									Effect: ↓OCD-VAS Response: Y-BOCS ↓≥35% from BL											7.2% (↓OCD-VAS)												Response: 0%					Intravenous infusion of ketamine (0.5 mg/kg) over 40 min					45.4% (↓OCD-VAS)			Response: 50%
Simpson et al. [111]	2013			Department of Psychiatry, Columbia University, New York State Psychiatric Institute, New York, NY, USA														100 patients on SSRI with still clinically significant OCD (Y-BOCS ≥ 16); 97 patients toN=38 Risperidone (up to 4 mg/d)N=40 EX/RP (17 session, 2 wkly)N=19 placebo										RCT comparing SSRI augmentation with either EX/RP therapy, risperidone (→ max 4.0 mg/d), or pill placebo × 8 wk in two centres									Effect: % ↓Y-BOCS score from BLResponse: Y-BOCS ↓≥25% from BL											10.81% (↓Y-BOCS)												Responders: 15% (Y-BOCS ↓≥25%)					Add-on risperidone (up to 4 mg/day)EX/RP (17, 2×week, 90 min-sessions)					EX/RP: 52.2% (↓Y-BOCS). Risperidone: 13.4% (↓Y-BOCS)			EX/RP response: 80% Risperidone response: 22.5%
Park et al. [112]	2014			Department of Psychology, University of South Florida,Tampa, FL, USA														30 children and adolescents with OCD (CY-BOCS≥16) stable on psychotropic medication × ≥ 12 wks: 15 to D-cycloSer; 15 to placebo										DB RCT of ERP + D-cycloSer (25-50 mg) after last 7 sessions vs. ERP + placebo × 10 wk									Effect: %↑ Homework compliance (rated with a 7-point Likert scale ranging from 0 (“did not complete any assigned homework”) to 6 (“completed all homework and made efforts above and beyond assignments”); no criterion for response											4.7% (↑ homework compliance)																	Exposure and response prevention therapy (ERP) + D-cycloSer (25-50 mg depending on weight) after last 7 session					-6% (↑ homework compliance)
Grant et al. [113]	2014			Pediatrics and Developmental Neuroscience Branch, NIMH, National Institutes of Health, Bethesda, MD, USA														60 treatment-resistant children and adolescents (7-17 years, CY-BOCS≥20); 30 to riluzole, 30 to placebo										DB RCT of add-on riluzole (up to 100 mg/day) vs. placebo × 12 wk									Effect: % ↓CY-BOCS score from BL.%↓CGI-S from BL%↑ CGAS from BLResponse: CY-BOCS ↓≥30% from BL											22.9% (↓CY-BOCS)10.7% (↓CGI-S)18,8% (↑CGAS)												Response: 18%					Add-on riluzole starting from 10 mg/day and increased daily up to 100 mg/day					20.1% (↓CY-BOCS)9.8% (↓CGI-S)12.8% (↑ CGAS)			Response: 16%
Author(s)	Year			Affiliation/ Country														Population (N)										Design									Outcome Measures (Response Criteria)											Placebo Effect (%)									Placebo Responders (%)								Drug(s) or other Treatment than Placebo					Drug Effect (%)			Drug Responders (%)
Mataix-Cols et al. [114]	2014			King’s College London, Institute of Psychiatry, London, UK														27 children and adolescents with OCD (CY-BOCS≥16); 13 to D-cycloSer, 14 to placebo										DB RCT of Exposure and response prevention therapy (ERP) + D-cycloSer 50 mg after each session vs. ERP + placebo × 17 wk									Effect: % ↓CY-BOCS score from BL.Response: CY-BOCS ↓≥35% from BLRemission: CY-BOCS score ≤ 10											59.3% (↓CY-BOCS)									Response: 64.2% (CY-BOCS ↓≥35% from BL); Remission: 42.8% (CY-BOCS ≤10)								Exposure and response prevention therapy (ERP) + D-cycloSer 50 mg after each session					60% (↓CY-BOCS)			Response: 61.5% (CY-BOCS ↓≥35% from BL); Remission: 53.8% (CY-BOCS score ≤10)
Afshar et al. [115]	2014			Isfahan Psychosomatic Research Center, Isfahan University of Medical Sciences, Isfahan, Iran														31 Y-BOCS≥16 to-N=16 Topiramate (mean dose: 137.5 mg/day)-N=15 PlaceboAdd-on on current SSRIs										12-wk, double-blind, placebo-controlled									Treatment effects: ↓Y-BOCS score from BL; response (r): Y-BOCS ↓≥25% from BL after 12 weeks											8.33% (↓Y-BOCS) after 12 weeks									14.28% (Y-BOCS ↓≥25% from BL after 12 weeks)								Topiramate (range 100-200, mean dose: 137.5 mg/day), initial dose of 25 mg/day increased by 25 mg weekly to a maximum 200 mg/day					19.81% (↓Y-BOCS) after 12 weeks			53.84% (Y-BOCS ↓≥25% from BL to 12 weeks)
Sarris et al. [116]	2015			Department of Psychiatry, The Melbourne Clinic, The University of Melbourne, Melbourne, VIC, Australia														34 to:N=18 3g/day NAC or N=16 Placebo(Y-BOCS≥16)										16-wk, double-blind, placebo-controlled, randomised trial									Treatment effects: ↓Y-BOCS score from BL; response: Y-BOCS ↓≥35% after 16 weeks											19.5% (↓Y-BOCS);									27% (Y-BOCS ↓≥35% from BL)								NAC (1.5 g q 12 h)					21% (↓Y-BOCS)			20% (Y-BOCS ↓≥35% from BL)
Pittenger et al. [117]	2015			Departments of Psychiatry and Psychology, Yale Child Study Center, Interdepartmental Neuroscience Program, Yale University School of Medicine, New Haven, CT, USA														40 DR OCD patients (SSRI/Clomi non-responders): 20 to riluzole, 18 to placebo										DB RCT of add-on riluzole (50 mg/day) vs. placebo × 12 wk									Effect: % ↓Y-BOCS score from BL; Partial response: Y-BOCS ↓≥25% from BL; Response: Y-BOCS ↓≥35% from BL											11% (↓Y-BOCS)									Partial or complete response: 11%								Add-on riluzole 50 mg/day after 2-week placebo lead-in phase					15% (↓Y-BOCS)			Partial or completeresponse: 26,3%
Jahanbakhsh et al. [118]	2016			Pharmaceutical Research Center, Department of Clinical Pharmacy, School of Pharmacy, Mashhad University of Medical Sciences, Mashhad, Iran														30 OCD patients currently treated with SSRI; 15 to W. somnifera, 15 to placebo										DB RCT of add-on W. somnifera extract 120 mg/day vs. placebo × 6 wk									Effect: % ↓Y-BOCS score from BLno criterion for response											11.1% (↓Y-BOCS)																	W. somnifera extract 120 mg/day					46.2% (↓Y-BOCS)
Paydary et al. [119]	2016			Psychiatric Research Center, Roozbeh Psychiatric Hospital, Tehran University of Medical Sciences, Tehran, Iran														44 drug naïve OCD patients (Y-BOCS ≥21); 22 to NAC add-on, 22 to placebo add-on										Multicentre DB RCT of fluvoxamine 200 mg/day + NAC 2000 mg/day vs. fluvoxamine 200 mg/day + placebo × 10 wk									Effect: % ↓Y-BOCS score from BLResponse: Y-BOCS ↓≥35% from BLRemission: Y-BOCS ≤ 16											30.3 (↓Y-BOCS)									Response: 22.7%								Fluvoxamine 100 mg/day for × first 4 weeks → 200 mg/day; + NAC 1000 mg/day × 1st week → 2000 mg/day					39% (↓Y-BOCS)			Response: 54.5%
Author(s)	Year			Affiliation/ Country								Population (N)										Design									Outcome Measures (Response Criteria)											Placebo Effect (%)											Placebo Responders (%)										Drug(s) or other Treatment than Placebo					Drug Effect (%)					Drug Responders (%)
Khalkhali et al. [120]	2016			Department of Psychiatry, Guilan University of Medical Sciences, Rasht, Iran								53 DR to:-N=26 Lamotrigine-N=27 placeboY-BOCS≥21 with stable SRIs dosages × at least 3 months before the study										12-wk, double blind, placebo-controlled RCT of 100 mg/day add-on lamotrigine to SSRIs in DR OCD-patients									Treatment effects:↓Y-BOCS score from BL; response: Y-BOCS ↓>25% from BL(at week 12)											17.7% (↓Y-BOCS);																					Adjunctive lamotrigine -doses (fixed 100 mg/day) (25→100 mg/day during the first 4 weeks, increments of 25 mg/week)					32.5% (↓Y-BOCS);
Rutrick et al. [121]	2016			Adams Clinical Trials, Watertown, MA, USA								50 DR toN=24 placebo or7N=26 Mavoglurant (augmentation to SSRIs) (Y-BOCS≥16)										Multicentre, randomised, DB, add-on phase 2 study × 16 wk									Treatment effects: ↓Y-BOCS from BL; Response (Y-BOCS ↓≥25% from BL) at week 17											32.06% (↓Y-BOCS)											50% (Y-BOCS ↓≥25% from BL)										(4 week up-titration period → 12-weeks fixed-dose 200 mg Mavoglurant q 12 h → 3-weeks tapering-off					26.59% (↓Y-BOCS)					34.5% (Y-BOCS ↓≥25% from BL)
Feng et al. [122]	2016			Department of Psychiatry, Tongde Hospital of Zhejiang Province, Hangzhou, Zhejiang, China								360 OCD patients with Y-BOCS ≥ 16 randomised to GROUP A (N=120) GROUP B (N=120) or GROUP C (N=120)										SB RCT of TEAS with CBT + clomi (GROUP A) vs. TEAS with CBT + placebo (GROUP B) vs. sham TEAS with CBT + clomi (GROUP C) × 12 wk									Effect: % ↓Y-BOCS score from BL, Response: Y-BOCS ↓≥35% from BL, Remission: CY-BOCS score ≤ 12											GROUP B: 45% (↓Y-BOCS)GROUP C: 38.2% (↓Y-BOCS)											GROUP B: Response: 82.5% Remission: 22.5% GROUP C: Response: 67.5% Remission: 9.2%										GROUP A: Transcutaneous electrical acupoint stimulation combined with CBT + clomi					GROUP A: 59.3% (↓Y-BOCS)					GROUP A: Response: 89.2%Remission: 29.2%
de Leeuw et al. [123]	2017			Altrecht Academic Anxiety Center, Utrecht, The Netherlands								39 patients with OCD randomised to D-cycloSer (N=19) or placebo (N=20) in add-on										DB RCT of Exposure and response prevention therapy (ERP) + D-cycloSer 125 mg/day vs. ERP + placebo × 8 wk									Effect: % ↓Y-BOCS score from BL.Partial response: Y-BOCS ↓≥25% from BLResponse: Y-BOCS ↓≥30% from BL											17% (↓Y-BOCS)											Partial or complete response: 35%										Exposure and response prevention therapy (ERP) + D-cycloSer 125 mg/day					25.2% (↓Y-BOCS)					Partial or complete response: 90%
Asnaani et al. [124]	2017			Department of Psychiatry, Center for the Treatment and Study of Anxiety, University of Pennsylvania, Philadelphia, PA, USA								100 patients on SRI with Y-BOCS ≥ 16 randomised to risperidone (N=40), EX/PR (N=40) or placebo (N=20)										RCT comparing EX/RP therapy + SSRI, risperidone (0.25 mg/day × 3 days, 0.5 mg/day × 4 days, → ↑0.5 mg/wk to max4.0 mg/day +SSRI vs. pill placebo + SSRI × 8 wk									Effect: ↑ QLESQ-SF score from BL; ↓SAS-SR and ↓SDS score from BL. No response criteria											11.6% (↑ QLESQ-SF); 5.5% (↓ SAS-SR); 19% (↓ SDS)																					SSRI augmentation with either EX/RP therapy (17 twice-weekly, 90-min sessions) or risperidone (0.25 mg/d × 3 days, 0.5 mg/d × 4 days, → ↑0.5 mg/week to max 4.0 mg/day)					EX/RP therapy: 20.6% (↑ QLESQ-SF); 14.8% (↓ SAS-SR); 51.5% (↓ SDS)Risperidone: 5.7% (↑ QLESQ-SF) 4.5% (↓ SAS-SR) 20.6% (↓ SDS)
Author(s)	Year			Affiliation/ Country								Population (N)													Design									Outcome Measures (Response Criteria)										Placebo Effect (%)									Placebo Responders (%)										Drug(s) or other Treatment than Placebo					Drug Effect (%)					Drug Responders (%)
Ahmadpanah et al.[125]	2017			Hamadan University of Medical Sciences (HUMS), Behavioral Disorders and Substance Abuse Research Center, Hamadan, Iran								43 DR OCD patients (not responding to SSRI or Clomi) to buprenorphine (N=23) or placebo (N=20)													DB RCT of DR patients with stabilised SSRI or clomi to add-on buprenorphine or placebo × 12 wk									Effect: % ↓Y-BOCS score from BL.Partial response: Y-BOCS ↓≥25% from BLResponse: Y-BOCS ↓≥35% from BL										7% (↓Y-BOCS)									Partial or complete response: 25%										Buprenorphine tablets (2-4 g; sublingual) daily					17,5% (↓Y-BOCS)					Partial or complete response: 39%
Modarresi et al. [126]	2017			Research Center for Rational Use of Drugs, Tehran University of Medical Sciences, Tehran, Iran								32 DR OCD patients (failing at least 3 adequate trials of antidepressant, including clomi) and Y-BOCS ≥ 24 randomised to memantine (N=16) or placebo (N=16)													DB RCT of add-on memantine 20 mg/day vs. placebo × 12 wk									Effect: % ↓Y-BOCS score from BL; Response: Y-BOCS ↓≥35% from BL										-0.2% (↓Y-BOCS)									Response: 0%										Add-on memantine 20 mg/day					31% (↓Y-BOCS)					Response: 73.3%
Costa et al. [127]	2017			Department & Institute of Psychiatry, University of São Paulo Medical School, São Paulo-SP, Brazil								40 DR OCD patients (not responding to SSRI or clomi) randomised to NAC (N=18) or placebo (N=22)													DB RCT of add-on NAC (up to 3000 mg/day) vs. placebo × 16 wk									Effect: % ↓Y-BOCS score from BLResponse: Y-BOCS ↓≥25% from BL										12.1% (↓Y-BOCS)									Response: 26.3%										Add-on NAC, 1,200 mg/day for the first week, 2,400 mg in the second week, 3,000 mg/day from the third week on					16.8% (↓Y-BOCS)					Response: 37.5%
Arabzadeh et al. [128]	2017			Psychiatric Research Center, Roozbeh Hospital, Tehran University of Medical Sciences, Tehran, Iran								44 drug naïve OCD patients (Y-BOCS ≥21) randomised to L-carnosine (N=22) or placebo (N=22)													DB RCT of fluvoxamine 100 - 200 mg/day + L‐carnosine 1000 mg/day vs. fluvoxamine 100 - 200 mg/day + placebo × 10 wk									Effect: % ↓Y-BOCS score from BLPartial response: Y-BOCS ↓≥25% from BL;Response: Y-BOCS ↓≥35% from BLRemission: Y-BOCS ≤ 14										24.3 (↓Y-BOCS)									Partial: 45.5%; Complete: 9.1%; Remission: 9.1%										Fluvoxamine 100 mg/day × 4 wk→200 mg/day × 6 wk + L‐carnosine 1000 mg/day					35.3 (↓Y-BOCS)					Partial: 45.5%Complete: 36.4%Remission: 27.3%

Abbreviations used: BDI=Beck Depression Inventory; BDZs=benzodiazepines; BL=baseline; BT= Behaviour Therapy; CBT=Cognitive-Behavioural Therapy; CGI=Clinical Global Impressions scale-i=improvement, s=severity; clomi=clomipramine; clonz=clonazepam; CCMD-2=Chinese Criteria of Mental Disorders, 2nd edition; CPRS=Comprehensive Psychopathological Rating Scale, Karolinska institutet; DB=double-blind; DLPFC=dorsolateral prefrontal cortex; DR=drug-resistant/treatment-resistant; D-cycloSer=D-cycloserine; EPA=eicosapentaenoic acid; ERP=exposure and ritual (response) prevention; GAF=Global Assessment of Functioning; GVC=gamma ventral capsulotomy; halo=haloperidol; Ham-D=Hamilton Depression Rating Scale; MOCI=Maudsley Obsessional Compulsive Inventory; MT=motor threshold; NAC=N-Acetylcysteine; NIMH= National Institute of Mental Health Global Obsessive-Compulsive Scale; OCS= Obsessive Compulsive Scale; PGI=Patient’s Global Impressions; RegSert=Standard sertraline dose; SloSert=slowly titrated sertraline; SROC=Self-Rating Obsessive-Compulsive Personality Inventory; SRON=Self-Rating Obsessional Neurotic Scale; wk=weeks; Y-BOCS=Yale-Brown Obsessive-Compulsive Scale; YGTSS, Yale Global Tick Severity Scale; ZnSO4=zinc sulfate; ↑=increase; ↓=reduction; *Patients were not DSM-IV OCD at inclusion, because they had responded to previous trial; hence, actual baseline was factitious and the differences observed were modest for all groups, meaning that treatment continued to work.