How Effective Are Second-Generation Antipsychotic Drugs?

Home , Quetiapine, Zotepine

Molecular Psychiatry (2009) 14, 429–447 & 2009 Nature Publishing Group All rights reserved 1359-4184/09 $32.00 www.nature.com/mp ORIGINAL ARTICLE How effective are second-generation antipsychotic drugs? A meta-analysis of placebo-controlled trials S Leucht1, D Arbter1, RR Engel2, W Kissling1 and JM Davis3 1Department of Psychiatry and Psychotherapy, Klinik fu¨r Psychiatrie und Psychotherapie der TU-Mu¨nchen, Klinikum rechts der Isar, Technische Universita¨tMu¨nchen, Mu¨nchen, Germany; 2Department of Psychiatry and Psychotherapy, Psychiatrische Klinik der Ludwig-Maximilian-Universita¨tMu¨nchen, Mu¨nchen, Germany and 3Department of Psychiatry, University of Illinois at Chicago, Chicago, IL, USA

We conducted a systematic review and meta-analysis of randomized controlled trials that compared second-generation antipsychotic (SGA) drugs with placebo in schizophrenic patients and which considered 13 different outcome measures. Thirty-eight randomized controlled trials with 7323 participants were included. All SGA drugs were more effective than placebo, but the pooled effect size (ES) for overall symptoms (primary outcome) was moderate (À0.51). The absolute difference (RD) in responder rates was at 18% (41% responded to drug compared with 24% to placebo, number needed to treat = 6). Similar ESs were found for the other efficacy parameters: negative symptoms (ES = À0.39), positive symptoms (ES = À0.48), depression (ES = À0.26), relapse (RD 20%) and discontinuation due to inefficacy (RD 17%). Curiously, the efficacy of haloperidol for negative and depressive symptoms was similar to that of the SGA drugs. In contrast to haloperidol, there was no difference in terms of EPS between any SGA drugs and placebo, and there was also no difference in terms of dropouts due to adverse events. Meta-regression showed a decline in treatment response over time, and a funnel plot suggested the possibility of publication bias. We conclude that the drug versus placebo difference of SGA drugs and haloperidol in recent trials was moderate, and that there is much room for more efficacious compounds. Whether methodological issues account in part for the relatively low efficacy ESs and the scarcity of adverse event differences compared with placebo needs to be established. Molecular Psychiatry (2009) 14, 429–447; doi:10.1038/sj.mp.4002136; published online 8 January 2008 Keywords: meta-analysis; schizophrenia; antipsychotic agents; treatment outcome; bias; methodology

Introduction has implications for the interpretation of comparisons between second-generation antipsychotic (SGA) Recent critics of psychotropic agents have claimed drugs and conventional antipsychotics. The review that these drugs are not efficacious. For example, the also assesses how well it can be documented that the efficacy of anticholinesterase inhibitors for Alzhei- newer drugs cause certain adverse effects. New versus mer’s dementia has been questioned,1 as has the old drug comparisons may establish that the new drug efficacy of modern antidepressants, where Moncrieff has a lower incidence of adverse effects, but they do and Kirsch2 found only a two-point difference not establish whether the newer drug can cause that between drug and placebo on the Hamilton rating adverse effect. Finally, the database allows for the scale for depression is found. In this context, we discussion of a number of design issues in the context present a meta-analysis of 38 randomly controlled of placebo-controlled research in schizophrenia. trials with 7323 participants comparing second- generation (atypical) antipsychotics with placebo. The aim is to assess the efficacy and safety of SGA Materials and methods drugs based on 13 outcomes. This large database Search allows for some judgments on the efficacy of anti- We searched the register of the Cochrane Schizophrenia psychotic drugs in general, and the degree of efficacy Group (CSG) for randomized controlled trials that compared oral routes of administration of SGAs (search Correspondence: Dr S Leucht, Klinik fu¨ r Psychiatrie und terms: amisulpride, aripiprazole, clozapine, olanza- Psychotherapie der TU-Mu¨nchen, Klinikum rechts der Isar, pine, quetiapine, risperidone, sertindole, ziprasidone Technische Universita¨tMu¨ nchen, Ismaningerstr. 22, Mu¨nchen and zotepine) with placebo and/or conventional anti- 81675, Germany. E-mail: [email protected] psychotics in the treatment of schizophrenia or related Received 29 March 2007; revised 21 November 2007; accepted 26 disorders (schizoaffective, schizophreniform or delu- November 2007; published online 8 January 2008 sional disorder, any diagnostic criteria). There were no Atypical antipsychotics versus placebo S Leucht et al 430 language restrictions. The last search was made in as a measure of acceptability of treatment. In a ‘once August 2005; since then, studies from monthly MED- randomized–analyzed’ approach, we assumed in the LINE searches until September 2006 were added. case of dichotomous data that participants who The CSG register is compiled by regular methodical dropped out prior to completion had no change in searches in numerous electronic databases (BIOSIS, their condition. CINAHL, Dissertation abstracts, EMBASE, LILACS, MEDLINE, PSYNDEX, PsycINFO, RUSSMED and Meta-analytic calculations Sociofile), supplemented by the hand searching of Standardized mean differences (SMDs) based on relevant journals and numerous conference proceed- Hedges’s adjusted g and its 95% confidence intervals ings (for details see the description of the Cochrane (CIs) were calculated for continuous data. When s.d. Schizophrenia Group3). We also searched the FDA were not reported, we either derived them from other web site and previous reviews4,5 including those of measures of variability or P-values, or we used the the Cochrane Collaboration. Only studies meeting average s.d. of the other studies. For dichotomous the quality criteria A (adequate randomization) and B data, relative risks (RRs) and risk differences (RDs) (usually studies stated to be randomized without further along with their 95% CIs were calculated. We believe details) according to the Cochrane handbook were that both measures are important. The mathematical included.6 We used only optimum doses of SGA drugs properties of RR are somewhat better than those of in fixed-dose studies as determined in controlled dose- RD, because they make an adjustment for base- finding studies as follows: amisulpride 50–300 mg dayÀ1 line risks.13 But RRs are often misinterpreted by for predominantly negative symptoms and 400– clinicians.14,15 The number of patients needed to 800 mg dayÀ1 for positive symptoms, aripiprazole treat (NNT) or the number of participants needed to 10–30 mg day, olanzapine 10–20 mg dayÀ1, quetiapine harm were calculated as the inverse of the RD. We >250mgdayÀ1, risperidone 4–6 mg dayÀ1, sertindole also showed the percentages in each group, because 16–24 mg dayÀ1 and ziprasidone 120–160 mg dayÀ1.It we feel that this is crucial for the reader to be able should be noted that there is a debate about the to appreciate the results. For example, a RR reduction optimum quetiapine doses, but there is no evidence of 50% is not meaningful if the reader does not know from dose-finding studies that shows higher doses whether these underlying percentages are 60 versus are more efficacious. Indeed, in the studies included 30% or 4 versus 2%. here the 750 mg quetiapine per day group was the We explored study heterogeneity by using the least effective one.7 Eleven studies had an additional I2 statistics, a measure estimating how much of the haloperidol arm. The results of the studies’ haloperidol variance is explained by study heterogeneity.16 Since groups as compared with placebo were also pooled as a in some of the analyses there was considerable benchmark. heterogeneity, we applied the random effects model by Der-Simonian and Laird17 throughout for the Data extraction and outcome parameters pooling of the studies. Random-effect models are in All data were extracted independently by two general more conservative than fixed-effect models, reviewers. The first authors (when addresses were because they take heterogeneity among studies into available) and all SGA drugs manufacturers were account. When studies had several arms (for example, contacted for missing data. The primary outcome of risperidone, quetiapine and placebo), we used the interest was the mean overall change of symptoms mean of the single arms to avoid counting the same according to the following hierarchy: the change of participants twice. the Positive and Negative Syndrome Scale (PANSS8) Unrestricted maximum likelihood random effects total score from baseline, if not available the change of meta-regression was used to find whether there was a the Brief Psychiatric Rating Scale (BPRS9), then change of the primary efficacy outcome (mean change values at study end point of these scales, all based of overall symptoms) over time using publication year on intent-to-treat data set whenever available. We also as a moderator. analyzed negative symptoms, positive symptoms, We made a sensitivity analysis excluding studies depressive symptoms and overall quality of life in a that consisted of patients with predominantly nega- similar fashion. For dichotomous efficacy measures, tive symptoms,18–23 long-term studies on initially we analyzed responder rates, relapse rates and stable patients24–28 and one very short study of only dropout due to inefficacy. The hierarchy for respon- 2 weeks duration.29 Owing to space limitations, we do der rates was 50% or more reduction from baseline on not show the results here, but any result that deviated the PANSS/BPRS or better; or a Clinical Global to an important extent from the primary analysis will Impression10 of much improved in so far as available; be mentioned. followed by the authors’ definitions, which were Studies with negative results are less likely to be usually at least 20 or 30% PANSS/BPRS reduction. published than studies with significant results. The Adverse effect outcomes were based on use of possibility of such publication bias was examined antiparkinson medication, mean EPS score (Simpson applying the ‘funnel plot’ method to the primary out- Angus Scale (SAS11), Extrapyramidal Symptoms Rat- come (mean change of overall symptoms) described ing Scale (ESRS12)), dropouts due to adverse events by Egger et al.30 All calculations were done with and sedation. Dropouts for any reason were analyzed Comprehensive Meta-Analysis Version 2.31 The exact

Molecular Psychiatry Table 1 Characteristics of included randomized controlled double-blind studies comparing second-generation antipsychotics with placebo (studies that provided comparisons for two second-generation antipsychotics are listed twice)

Study Antipsychotics and daily dose in n Duration (weeks)/ Mean duration Selected characteristics of patients mg, for flexible dose studies mean minimum of illness in value (range) washout (days) years

Amisulpride 18 AMI (100; 300), (34; 36), 6, (42) 11 Schizophrenia, disorganized, catatonic, undifferentiated or PBO 34 residual type, patients had to comply with Andreasen’s criteria for ‘negative schizophrenia’ (DSM-III) 19 AMI (50; 100), 159 12, (28) B10 Patients with residual schizophrenia and predominantly negative PBO 83 symptoms (DSM-III-R) 20 AMI 150, 70, 26, (2) NI Schizophrenia with predominantly negative symptoms OLA (5; 20), (70; 70), (criteria, NI) PBO 34 21 AMI 100, 69, 26, (0) 10 Patients with residual or hebephrenic schizophrenia and PBO 72 predominantly negative symptoms (DSM-III-R) 22 AMI 50, 14, 6, (8) 3 Schizophrenia and schizotypal personality disorder, young PBO 13 patients with mainly negative symptoms (DSM-III-R)

Aripiprazole 32 ARI (15; 30), (102; 102), 4, (5) 16.3 Schizophrenia or schizoaffective disorder with acute relapse HAL 10, 104, (DSM-IV)

PBO 106 Leucht S placebo versus antipsychotics Atypical 33 ARI (2; 5; 10), (93; 92; 94), 6, (NI) NI Hospitalized patients in acute relapse of schizophrenia (DSM-IV) PBO 88 24 ARI 15, 155, 26, (3) NI Stabilized patients with chronic schizophrenia (DSM-IV) al et PBO 155 34 ARI (20; 30), (101; 101), 4, (5) NI Schizophrenia or schizoaffective disorder with acute relapse PBO, 103 (DSM-IV) RIS 6 99 35 ARI (10; 15; 20), (106; 106; 6, (2) NI Schizophrenia with acute relapse (DSM-IV) PBO 100), 108 36 ARI 30 (5–30), 34, 4, (3) NI Schizophrenia with acute relapse (DSM-III-R) HAL 20 (5–20), 34, PBO 35 37 ARI (2; 10; 30), (59; 60; 61), 4, (3) NI Schizophrenia with acute relapse (DSM-IV) HAL 10, 63, PBO 64

Clozapine 38 CLO 608 (NI), 16, 4, (3) NI Acute exacerbation of schizophrenia (clinical diagnosis) PBO 8 oeua Psychiatry Molecular Olanzapine 39 OLA (5±2.5a,10±2.5, 15±2.5), (65; 64; 69), 6, (4) 14 Acute exacerbation of schizophrenia (DSM-III-R) HAL 16.4, PBO 69, 68 431 432 oeua Psychiatry Molecular

Table 1 Continued

40 OLA (1; 10), (52; 50), 6, (4) 16 Schizophrenia with an acute exacerbation (DSM-III-R) PBO 50 41 OLA NI (10–20), 224, 28, (0) 11 Healthy outpatients with schizophrenia or schizoaffective PBO 102 disorder (DSM-IV) 42 OLA 15, 93, 6, (3) 13 Schizophrenia (DSM-IV) PBO 87 43 OLA 12.5, 72, 7, (2) NI Schizophrenia (DSM-IV) PBO 35 placebo versus antipsychotics Atypical 20 OLA (5; 20), (70; 70), 26, (2) NI Schizophrenia with predominantly negative symptoms (criteria, AMI 150, 70, NI) PBO 34

Quetiapine 7 QUE (75; 150; 300; 600; 750), (53; 48; 52; 6, (7) 15 Acute exacerbation of (sub-) chronic schizophrenia (DSM-III-R) Leucht S HAL 12, PBO 51; 54), 52, 51 44 QUE 307 (75–750), 54, 6, (2) 15 Acute exacerbation of (sub-) chronic schizophrenia (DSM-III-R) al et PBO 55 45 QUE 250, 8, 3, (2) 12 (Sub-) chronic schizophrenia PBO 4 (DSM-III-R) 29 QUE 484, 156, 2, (NI) NI Acute exacerbation of schizophrenia (DSM-IV) or schizoaffective RIS, 153, disorder (MINI-PLUS) PBO 73 46 QUE 209 ( < 250), 94, 6, (1) 15 Acute exacerbation of (sub-) chronic schizophrenia (DSM-III-R) QUE 360 ( > 250), 96, PBO 96

Risperidone 25 RIS NI (2–6), 22, 50.2, (28) NI Chronic schizophrenia (DSM-IV) PBO 20 47 RIS 7.8 (2–10), 53, 6, (7) 15 Schizophrenia (DSM-III-R) HAL 15.0 (4–20), 53, PBO 54 48 RIS (2; 6; 10; 16), (24; 22; 22; 8, (3) 16 Chronic schizophrenia (DSM-III-R) HAL 20, PBO 24), 21, 22 49 RIS (2; 6; 10; 16), (63; 64; 65; 8, (3) 16 Chronic schizophrenia (DSM-III-R) HAL 20, PBO 64), 66, 66 34 RIS 6, 99, (101; 4, (5) NI Schizophrenia or schizoaffective disorder with acute relapse ARI (20; 30), 101), (DSM-IV) PBO 103 Table 1 Continued

29 RIS, 153, 2, (NI) NI Acute exacerbation of schizophrenia (DSM-IV) or schizoaffective QUE 484, 156, disorder (MINI-PLUS) PBO 73 50 RIS 4, 85, 4, (NI) NI Chronic schizophrenia (DSM-III-R) RIS 8, 78, PBO 83

Sertindole 51 SER (8; 12; 20), (52; 51; 54), 6, (4) 14 Schizophrenia, history of a previous response to antipsychotic PBO 48 drugs (DSM-III-R) 52 SER (20; 24), (117; 113), 8, (4) 15 Schizophrenia (DSM-III-R or DSM-IV) HAL (16), 115, PBO 116 53 SER (12; 20; 24), (76; 68; 72), 8, (4) 16 Schizophrenia, history of a previous response to antipsychotic HAL (4; 8; 16), (71; 67; 70), drugs (DSM-III-R or DSM-IV) PBO 73

Ziprasidone 26 ZIP (40; 80; 160), (76; 72; 71), 52, (0) NI Chronic, stable, in-patients with schizophrenia (DSM-III-R)

PBO 75 Leucht S placebo versus antipsychotics Atypical 54 ZIP (80; 160), (106; 104), 6, (3) 14 Acute exacerbation of (sub-) chronic schizophrenia or

PBO 92 schizoaffective disorder (DSM-III-R) al et 55 ZIP (40; 120), (44; 47), 4, (4) 16 Acute exacerbation of schizophrenia or schizoaffective disorder PBO 48 (DSM-III-R) 56 ZIP (40; 120; 200), (87; 78; 86), 6, (3) NI Acute exacerbation of schizophrenia or schizoaffective disorder HAL 15, 85, (DSM-III-R) PBO 83

Zotepine 28 ZOT 241 (150 or 300), 53, 8, (0) 11 Acute exacerbation of (sub-) chronic schizophrenia (DSM-III-R CHLOR 532 (300 or 600), 53, PBO 53 27 ZOT NI (150–300), 63, 26, (0) 14 Chronic schizophrenia (DSM-III-R) PBO 58 23 ZOT 131 (25–225), 39, 8, (B5) 11 Residual schizophrenia with stable primary negative symptoms PBO 46 (ICD-10)

Abbreviations: AMI, amisulpride; ARI, aripiprazole; CLO, clozapine; DSM-III, -III-R, -IV, different versions of the Diagnostic and Statistical Manual of Mental Disorders; HAL, haloperidol; ICD-9/ICD-10, International Classification of Diseases, 9th/10th Revision; MDI, mean duration of illness (years); n, number of patients; NI, not oeua Psychiatry Molecular indicated; OLA, olanzapine; PBO, placebo; QUE, quetiapine; RIS, risperidone; SER, sertindole; ZIP, ziprasidone; ZOT, zotepine. aMarks dosage groups that were excluded because they lay outside the optimum dose according to randomized dose finding studies. 433 Atypical antipsychotics versus placebo S Leucht et al 434 formulae were reported there. Two-sided P-values sertindole (N = 3), ziprasidone (N = 4), zotepine (N =3; < 0.05 were considered statistically significant. three studies provided results on two SGA drugs). Most of the studies were short-term and examined patients with positive symptoms, while only six Results studies examined patients with predominantly negative symptoms (four amisulpride studies, one olanza- The search pine and amisulpride study and one zotepine study). The searches in the register of the Cochrane Schizo- Almost all studies were conducted by pharmaceutical phrenia Group yielded 4166 citations. Of those companies and usually for registrational purposes. publications that we ordered for inspection, 107 The minimum duration of washout was usually not studies were excluded for the following reasons: no more than a few days. The median of mean age was or inadequate randomization (N = 50), no appropriate 38 years (see Table 1). intervention or control group (N = 29), inappropriate participants (N = 2), no usable data (N = 24), presenta- Outcomes tion of only a subgroup (N = 1) and very short duration The results in terms of SMDs or RRs for the single (5 days, N = 1). The results of 202 studies comparing drugs are shown in Figures 1–11. Table 2 presents SGAs with first-generation antipsychotics will be pooled results of the single drugs based on RDs and reported elsewhere. Thirty-eight studies with 7323 NNTs. Table 3 presents all results for dropout rates. participants were included (only the principle publication of each study is referenced): amisulpride Overall efficacy (N = 5), aripiprazole (N = 7), clozapine (N = 1), olanza- All antipsychotics were significantly more efficacious pine (N = 6), quetiapine (N = 5), risperidone (N = 7), than placebo in the treatment of overall symptoms

Study name Comparison Statistics for each study Hedges's g and 95% CI Hedges's Lower Upper glimit limit p-Value Total Danion 1999 AMI -0.54 -0.81 -0.27 0.0001 241 Amisulpride pooled -0.54 -0.81 -0.27 0.0001 241 Kane 2002 ARI -0.43 -0.67 -0.19 0.0005 301 Modell 2005 ARI -0.33 -0.62 -0.03 0.0297 179 Pigott 2003 ARI -0.30 -0.53 -0.07 0.0094 293 Potkin 2003 ARI -0.44 -0.68 -0.20 0.0003 297 Study 138001 2002 ARI -0.43 -0.67 -0.19 0.0004 307 Study 93202 2002 ARI -0.64 -1.12 -0.16 0.0094 68 Study 94202 2002 ARI -0.53 -0.90 -0.15 0.0060 111 Aripiprazole pooled -0.41 -0.51 -0.31 0.0000 1556 Honigfeld 1984a CLO -1.64 -2.61 -0.68 0.0009 22 Clozapine pooled -1.64 -2.61 -0.68 0.0009 22 Arvanitis 1997 HAL -0.62 -1.02 -0.23 0.0020 101 Beasley 1996 HAL -0.63 -0.98 -0.28 0.0005 130 Borison 1992 HAL -0.51 -0.89 -0.12 0.0095 107 Chouinard 1993 HAL -0.59 -1.19 0.01 0.0539 43 Kane 2002 HAL -0.46 -0.74 -0.18 0.0013 201 Marder 1994 HAL -0.35 -0.70 -0.01 0.0463 128 Study 115 2000 HAL -0.58 -0.89 -0.26 0.0003 162 Study 93202 2002 HAL -0.91 -1.40 -0.41 0.0003 68 Study 94202 2002 HAL -0.37 -0.74 0.01 0.0536 111 Zborowski 1995 HAL -0.52 -0.78 -0.25 0.0002 219 Zimbroff 1997 HAL -0.58 -0.85 -0.30 0.0000 270 Haloperidol pooled -0.53 -0.64 -0.43 0.0000 1540 Beasley 1996 OLA -0.65 -0.96 -0.34 0.0000 189 Beasley 1996a OLA -0.71 -1.11 -0.30 0.0006 98 Beasley 2003 OLA -0.22 -0.46 0.01 0.0613 326 Corrigan 2004 OLA -0.96 -1.27 -0.65 0.0000 178 Kryzhanovskaya 2006 OLA -0.59 -1.00 -0.18 0.0049 107 Lecrubier 1999 OLA -0.44 -0.88 -0.01 0.0466 94 Olanzapine pooled -0.59 -0.83 -0.35 0.0000 992 Arvanitis 1997 QUE -0.62 -0.94 -0.30 0.0001 206 Borison 1996 QUE -0.36 -0.74 0.02 0.0627 106 Fabre 1995 QUE -1.47 -2.73 -0.22 0.0213 12 Potkin 2006 QUE -0.02 -0.30 0.26 0.9082 227 Small 1997 QUE -0.49 -0.78 -0.20 0.0010 184 Quetiapine pooled -0.42 -0.72 -0.13 0.0051 735 Bai 2003 RIS -0.42 -1.02 0.18 0.1677 42 Borison 1992 RIS -0.66 -1.04 -0.27 0.0009 107 Chouinard 1993 RIS -1.22 -1.85 -0.59 0.0002 44 Marder 1994 RIS -0.93 -1.29 -0.56 0.0000 127 Potkin 2003 RIS -0.49 -0.77 -0.21 0.0006 198 Potkin 2006 RIS -0.42 -0.70 -0.13 0.0041 223 Study Ris-USA-72 1996 RIS -0.37 -0.64 -0.10 0.0078 236 Risperidone pooled -0.59 -0.78 -0.39 0.0000 977 van Kammen1996 SER -0.35 -0.75 0.04 0.0800 98 Zborowski 1995 SER -0.35 -0.58 -0.11 0.0037 325 Zimbroff 1997 SER -0.56 -0.85 -0.27 0.0002 206 Sertindole pooled -0.42 -0.58 -0.25 0.0000 629 Arato 2002 ZIP -0.58 -0.91 -0.25 0.0006 146 Daniel 1999 ZIP -0.51 -0.79 -0.22 0.0005 194 Keck 1998 ZIP -0.47 -0.89 -0.05 0.0283 88 Study 115 2000 ZIP -0.38 -0.69 -0.06 0.0188 156 Ziprasidone pooled -0.48 -0.65 -0.32 0.0000 584 Cooper 2000a ZOT -0.88 -1.28 -0.49 0.0000 106 Cooper 2000b ZOT -0.46 -0.83 -0.10 0.0120 119 Möller 2004 ZOT -0.29 -0.73 0.15 0.1977 79 Zotepine pooled -0.55 -0.89 -0.21 0.0013 304 -2.00 -1.00 0.00 1.00 2.00 Favours Antipsychotic Favours Placebo

Figure 1 Antipsychotic drugs versus placebo—Positive and Negative Syndrome Scale (PANSS)/Brief Psychiatric Rating Scale (BPRS) total score. AMI, amisulpride; ARI, aripiprazole; CLO, clozapine; HAL, haloperidol; OLA, olanzapine; QUE, quetiapine; RIS, risperidone; SER, sertindole; ZIP, ziprasidone; ZOT, zotepine; circles are effect sizes of single studies; diamonds of pooled results.

Molecular Psychiatry Atypical antipsychotics versus placebo S Leucht et al 435 Study name Comparison Statistics for each study Risk ratio and 95% CI Risk Lower Upper ratio limit limit p-Value Danion 1999 AMI 0.62 0.52 0.76 0.0000 Lecrubier 1999 AMI 0.81 0.53 1.23 0.3202 Loo 1997 AMI 0.68 0.55 0.86 0.0009 Amisulpride pooled 0.66 0.58 0.76 0.0000 Kane 2002 ARI 0.82 0.73 0.93 0.0021 Potkin 2003 ARI 0.83 0.71 0.96 0.0122 Study 138001 2002 ARI 0.82 0.70 0.96 0.0126 Study 93202 2002 ARI 0.74 0.53 1.02 0.0648 Study 94202 2002 ARI 0.76 0.60 0.96 0.0240 Aripiprazole pooled 0.81 0.75 0.87 0.0000 Arvanitis 1997 HAL 0.80 0.58 1.12 0.1923 Beasley 1996 HAL 0.78 0.59 1.02 0.0663 Borison 1992 HAL 0.70 0.50 0.96 0.0293 Chouinard 1993 HAL 0.61 0.39 0.94 0.0260 Kane 2002 HAL 0.89 0.78 1.03 0.1107 Marder 1994 HAL 0.90 0.74 1.10 0.3169 Study 93202 2002 HAL 0.59 0.40 0.87 0.0082 Study 94202 2002 HAL 0.86 0.70 1.06 0.1463 Zborowski 1995 HAL 0.96 0.87 1.06 0.4199 Zimbroff 1997 HAL 0.90 0.82 0.99 0.0273 Haloperidol pooled 0.86 0.80 0.93 0.0001 Beasley 1996 OLA 0.82 0.66 1.01 0.0675 Beasley 1996a OLA 0.83 0.69 0.98 0.0333 Kryzhanovskaya 2006 OLA 0.84 0.65 1.10 0.2007 Lecrubier 1999 OLA 0.72 0.49 1.05 0.0848 Olanzapine pooled 0.82 0.73 0.92 0.0005 Arvanitis 1997 QUE 0.81 0.63 1.04 0.0956 Borison 1996 QUE 0.95 0.76 1.18 0.6214 Fabre 1995 QUE 0.11 0.01 1.89 0.1284 Potkin 2006 QUE 1.04 0.84 1.29 0.7223 Small 1997 QUE 0.76 0.59 0.98 0.0314 Quetiapine pooled 0.88 0.75 1.04 0.1252 Bai 2003 RIS 0.45 0.23 0.89 0.0222 Borison 1992 RIS 0.54 0.36 0.79 0.0016 Chouinard 1993 RIS 0.32 0.16 0.64 0.0013 Marder 1994 RIS 0.56 0.41 0.75 0.0002 Potkin 2003 RIS 0.80 0.67 0.97 0.0227 Potkin 2006 RIS 0.81 0.63 1.03 0.0794 Study Ris-USA-72 1996 RIS 0.59 0.44 0.79 0.0004 Risperidone pooled 0.62 0.51 0.75 0.0000 van Kammen 1996 SER 0.91 0.76 1.09 0.3193 Zborowski 1995 SER 0.98 0.91 1.07 0.7008 Zimbroff 1997 SER 0.83 0.74 0.93 0.0018 Sertindole pooled 0.91 0.81 1.02 0.1167 Daniel 1999 ZIP 0.84 0.72 0.98 0.0292 Keck 1998 ZIP 0.77 0.57 1.03 0.0768 Ziprasidone pooled 0.82 0.71 0.94 0.0058 Cooper 2000a ZOT 0.44 0.27 0.72 0.0012 Cooper 2000b ZOT 0.90 0.78 1.05 0.1771 Zotepine pooled 0.65 0.32 1.33 0.2394 0.1 0.2 0.5 1 2 5 10 Favours Antipsychotic Favours Placebo

Figure 2 Antipsychotic drugs versus placebo—non-responder rates. AMI, amisulpride; ARI, aripiprazole; CLO, clozapine; HAL, haloperidol; OLA, olanzapine; QUE, quetiapine; RIS, risperidone; SER, sertindole; ZIP, ziprasidone; ZOT, zotepine; circles are effect sizes of single studies, diamonds of pooled results.

(primary outcome). Nevertheless, with the exception d.f. = 33, P < 0.001, see Figure 3). The meta-regression of clozapine (SMD = –1.65, based on only one study), with publication year as a moderator suggested the effect sizes (ESs) were moderate (pooled ES of the drug–placebo difference may have become smal- all SGA drugs: N = 35, n = 5568, SMD = À0.51, CI: ler over time (see Figure 4). This effect was no longer À0.58 to À0.43, P < 0.0001). This point is underscored statistically significant in the sensitivity analysis by the difference in responder rates. Sertindole was excluding patients with predominantly negative not significantly more efficacious than placebo, and symptoms and long-term or very short-term studies. quetiapine and zotepine were only significant in It should be noted that the subset in the sensitivity the sensitivity analysis (see Figure 2). The pooled analysis was more homogeneous, but statistical power RR across SGA drugs was 0.78 (CI: 0.73–0.83, N = 28, was also reduced (see Figures 1, 2 and 4; Table 2). n = 4498, P < 0.0001), and the associated RD was –0.18 Seven studies on relapse of 6–12 months duration (CI: À0.22 to À0.14, n = 4498, P < 0.0001), thus reflect- showed that aripiprazole, olanzapine, ziprasidone ing an 18% difference in responder rates (overall 41 and zotepine reduced the relapse risk significantly versus 24% responded under SGA drugs and placebo, more than placebo. The RR for relapse suggested a respectively) or an NNT of 6 (CI: 5–7). The sensitivity more pronounced superiority of the SGA drugs than analysis found an almost identical RR (0.79) and RD the RR for responder rates, but the RD was similar (À0.17). The funnel plot was asymmetrical, raising (all SGA drugs combined: N =7, n = 1371, RR = 0.41, the possibility that studies with negative results have CI: 0.28 to 0.59, RD = À0.20, CI: À11 to À30, NNT = 5, not been published (Egger’s regression intercept, CI: 3–9, P < 0.0001). Amisulpride was not superior to

Molecular Psychiatry Atypical antipsychotics versus placebo S Leucht et al 436 placebo. Data on clozapine, risperidone and sertin- The pooled ES across SGA drugs was: N = 30, n = dole were not available (see Figure 5). 4941, SMD = À0.48, CI: À0.57 to À0.38, P < 0.0001 (see Figure 6). Positive symptoms Amisulpride and zotepine showed no difference in Negative symptoms positive symptoms compared with placebo, but for In contrast to clozapine (only one study) and both drugs only studies on patients with predomi- quetiapine (P = 0.07, the effect was significant in the nantly negative symptoms were available. While there sensitivity analysis excluding the only 2-week study), were no data on clozapine, the other antipsychotics the other antipsychotics improved negative symp- were significantly more effective than placebo in the toms more than placebo. The ESs for negative symp- treatment of positive symptoms, with ESs ranging toms were usually lower than those for positive between À0.36 (aripiprazole) and À0.82 (risperidone). symptoms (ES across SGA drugs: N = 36, n = 5403, SMD = À0.39, CI: À0.45 to À0.33, P < 0.0001). It should be noted that most of the studies investigated patients with predominantly positive symptoms. Funnel Plot of Standard Error by Hedges's g 0.0 Amisulpride is the only compound for which several studies on patients suffering predominantly from negative symptoms are available. Such populations 0.2 are more appropriate for examining effects on negative symptoms. One such study showed no superiority of zotepine.23 In one such olanzapine study, the 0.4 5 mg dayÀ1 group was effective, but the 20 mg dayÀ1 group was not.20 Haloperidol also reduced overall negative symptoms significantly more than placebo Standard Error 0.6 (see Figure 7).

Depressive symptoms 0.8 On the basis of more limited data (14 studies), -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 the SGA drugs also reduced depressive symptoms Hedges's g more than placebo (N = 14, n = 1910, SMD = À0.26, CI: Figure 3 Funnel plot Positive and Negative Syndrome À0.38 to À0.15, P < 0.0001). Amisulpride, haloperi- Scale (PANSS)/Brief Psychiatric Rating Scale (BPRS) total dol, olanzapine, ziprasidone and zotepine were found score. The haloperidol arms were excluded from the funnel statistically significantly superior to placebo. Halo- plot. Egger’s regression intercept suggested statistically peridol also significantly reduced depression scores significant asymmetry (d.f. = 33, P < 0.001). (see Figure 8).

Figure 4 Meta-regression on the effects of publication year on the effect size for the difference between second-generation antipsychotic (SGA) drugs and placebo on reduction of overall symptoms. Slope = 0.02, Q = 6.83, d.f. = 1, P = 0.0090. Circle size reflects the weight a study obtained in the meta-regression. Note that excluding the outlier on the left (an early clozapine study by Honigfeld et al.38 did not reverse statistical significance (slope = 0.02, Q = 4.39, d.f. = 1, P = 0.0362), but using only those studies included in the sensitivity analysis did (slope = 0.01, Q = 1.44, d.f. = 1, P = 0.2295).

Molecular Psychiatry Atypical antipsychotics versus placebo S Leucht et al 437 Study name Comparison Statistics for each study Risk ratio and 95% CI Risk Lower Upper p-Value Total ratio limit limit Loo 1997 AMI 0.83 0.23 2.98 0.7809 141 Lecrubier 2006 AMI 0.39 0.16 0.96 0.0398 105 Amisulpride pooled 0.50 0.24 1.05 0.0660 246 Pigott 2003 ARI 0.59 0.45 0.77 0.0001 310 Aripiprazole pooled 0.59 0.45 0.77 0.0001 310 Beasley 2000 OLA 0.15 0.07 0.30 0.0000 326 Beasley 1996 OLA 0.41 0.20 0.87 0.0196 58 Lecrubier 2006 OLA 0.42 0.20 0.87 0.0202 175 Olanzapine pooled 0.29 0.15 0.58 0.0004 559 Arato 2002 ZIP 0.57 0.44 0.74 0.0000 277 Ziprasidone pooled 0.57 0.44 0.74 0.0000 277 Cooper 2000 ZOT 0.18 0.07 0.50 0.0009 119 Zotepine pooled 0.18 0.07 0.50 0.0009 119 0.01 0.1 1 10 100 Favours Antipsychotics Favours Placebo

Figure 5 Antipsychotics versus placebo—relapse. AMI, amisulpride; ARI, aripiprazole; OLA, olanzapine; ZIP, ziprasidone; ZOT, zotepine; circles are effect sizes of single studies, diamonds of pooled results.

Study name Comparison Statistics for each study Hedges's g and 95%CI Hedges's Lower Upper g limit limit p-Value Total Danion 1999 AMI -0.46 -0.73 -0.19 0.0008 241 Loo 1997 AMI 0.07 -0.26 0.40 0.6667 141 Pallière-Martinot 1995 AMI 0.05 -0.79 0.89 0.8997 20 Amisulpride pooled -0.16 -0.58 0.26 0.4569 402 Kane 2002 ARI -0.44 -0.68 -0.20 0.0003 302 Modell 2005 ARI -0.31 -0.60 -0.01 0.0398 179 Pigott 2003 ARI -0.30 -0.53 -0.07 0.0094 293 Potkin 2003 ARI -0.40 -0.64 -0.16 0.0011 297 Study 138001 2002 ARI -0.39 -0.63 -0.16 0.0011 307 Study 94202 2002 ARI -0.17 -0.54 0.20 0.3623 111 Aripiprazole pooled -0.36 -0.46 -0.25 0.0000 1489 Arvanitis 1997 HAL -0.63 -1.03 -0.24 0.0018 101 Beasley 1996 HAL -0.58 -0.92 -0.23 0.0012 131 Chouinard 1993 HAL -0.86 -1.48 -0.25 0.0059 43 Kane 2002 HAL -0.41 -0.68 -0.13 0.0042 202 Marder 1994 HAL -0.52 -0.87 -0.17 0.0034 128 Study 115 2000 HAL -0.58 -0.89 -0.27 0.0003 162 Study 94202 2002 HAL -0.23 -0.61 0.14 0.2157 111 Zborowski 1995 HAL -0.64 -0.91 -0.37 0.0000 219 Zimbroff 1997 HAL -0.56 -0.84 -0.29 0.0001 270 Haloperidol pooled -0.54 -0.65 -0.43 0.0000 1367 Beasley 1996 OLA -0.53 -0.84 -0.23 0.0006 191 Beasley 1996a OLA -0.67 -1.08 -0.27 0.0011 98 Corrigan 2004 OLA -0.50 -0.80 -0.21 0.0009 178 Kryzhanovskaya 2006 OLA -0.65 -1.06 -0.24 0.0020 107 Lecrubier 1999 OLA -0.54 -0.98 -0.10 0.0155 94 Olanzapine pooled -0.56 -0.72 -0.41 0.0000 668 Arvanitis 1997 QUE -0.63 -0.95 -0.31 0.0001 206 Borison 1996 QUE -0.39 -0.77 -0.01 0.0455 106 Fabre 1995 QUE -1.53 -2.80 -0.27 0.0177 12 Potkin 2006 QUE -0.10 -0.38 0.18 0.4948 227 Small 1997 QUE -0.40 -0.69 -0.11 0.0072 184 Quetiapine pooled -0.41 -0.67 -0.16 0.0015 735 Chouinard 1993 RIS -1.31 -1.95 -0.67 0.0001 44 Marder 1994 RIS -0.85 -1.21 -0.49 0.0000 127 Potkin 2003 RIS -0.49 -0.77 -0.21 0.0006 198 Potkin 2006 RIS -0.56 -0.84 -0.27 0.0001 223 Study Ris-USA-72 1996 RIS -1.15 -1.44 -0.86 0.0000 236 Risperidone pooled -0.83 -1.13 -0.53 0.0000 828 van Kammen1996 SER -0.30 -0.69 0.10 0.1420 98 Zborowski 1995 SER -0.33 -0.56 -0.10 0.0057 325 Zimbroff 1997 SER -0.50 -0.79 -0.21 0.0007 206 Sertindole pooled -0.38 -0.54 -0.21 0.0000 629 Arato 2002 ZIP -0.48 -0.81 -0.16 0.0038 146 Daniel 1999 ZIP -0.85 -1.14 -0.55 0.0000 194 Keck 1998 ZIP -0.41 -0.83 0.01 0.0575 88 Study 115 2000 ZIP -0.29 -0.60 0.03 0.0723 156 Ziprasidone pooled -0.52 -0.78 -0.26 0.0001 584 Möller 2004 ZOT -0.27 -0.71 0.17 0.2338 79 Zotepine pooled -0.27 -0.71 0.17 0.2338 79 -2.00 -1.00 0.00 1.00 2.00 Favours Antipsychotic Favours Placebo

Figure 6 Antipsychotic drugs versus placebo—positive symptoms. AMI, amisulpride; ARI, aripiprazole; CLO, clozapine; HAL, haloperidol; OLA, olanzapine; QUE, quetiapine; RIS, risperidone; SER, sertindole; ZIP, ziprasidone; ZOT, zotepine; circles are effect sizes of single studies, diamonds of pooled results.

Molecular Psychiatry Atypical antipsychotics versus placebo S Leucht et al 438 Study name Comparison Statistics for each study Hedges's g and 95% CI Hedges's Lower Upper p-Value Total g limit limit Boyer 1995 AMI -0.57 -0.99 -0.16 0.0068 104 Danion 1999 AMI -0.56 -0.83 -0.29 0.0000 241 Lecrubier 1999 AMI -0.22 -0.65 0.21 0.3081 97 Loo 1997 AMI -0.69 -1.02 -0.35 0.0001 141 Palliere-Martinot 1995 AMI -0.94 -1.83 -0.05 0.0378 20 Amisulpride pooled -0.56 -0.73 -0.39 0.0000 603 Kane 2002 ARI -0.28 -0.52 -0.04 0.0208 301 Modell 2005 ARI -0.27 -0.56 0.03 0.0757 179 Pigott 2003 ARI -0.17 -0.40 0.06 0.1386 293 Potkin 2003 ARI -0.39 -0.63 -0.15 0.0016 297 Study 138001 2002 ARI -0.44 -0.68 -0.21 0.0003 307 Study 93202 2002 ARI -0.44 -0.92 0.03 0.0693 68 Study 94202 2002 ARI -0.42 -0.79 -0.05 0.0272 111 Aripiprazole pooled -0.33 -0.43 -0.22 0.0000 1556 Honigfeld 1984a CLO -0.42 -1.27 0.42 0.3259 22 Clozapine pooled -0.42 -1.27 0.42 0.3259 22 Arvanitis 1997 HAL -0.71 -1.11 -0.31 0.0005 100 Beasley 1996 HAL -0.28 -0.62 0.06 0.1005 133 Borison 1992 HAL -0.37 -0.75 0.01 0.0587 107 Chouinard 1993 HAL -0.28 -0.87 0.31 0.3495 43 Kane 2002 HAL -0.28 -0.56 -0.00 0.0467 201 Marder 1994 HAL -0.10 -0.44 0.25 0.5839 128 Study 115 2000 HAL -0.43 -0.74 -0.12 0.0065 162 Study 93202 2002 HAL -0.56 -1.04 -0.08 0.0227 68 Study 94202 2002 HAL -0.17 -0.54 0.20 0.3726 111 Zborowski 1995 HAL -0.11 -0.37 0.16 0.4202 219 Zimbroff 1997 HAL -0.36 -0.64 -0.08 0.0107 255 Haloperidol pooled -0.30 -0.41 -0.20 0.0000 1527 Beasley 1996 OLA -0.45 -0.75 -0.15 0.0032 192 Beasley 1996a OLA -0.58 -0.98 -0.17 0.0049 98 Corrigan 2004 OLA -0.50 -0.80 -0.20 0.0010 178 Kryzhanovskaya 2006 OLA -0.36 -0.76 0.04 0.0803 107 Lecrubier 1999 OLA -0.27 -0.67 0.12 0.1783 158 Olanzapine pooled -0.44 -0.60 -0.29 0.0000 733 Arvanitis 1997 QUE -0.48 -0.80 -0.15 0.0038 196 Borison 1996 QUE -0.36 -0.74 0.02 0.0637 106 Fabre 1995 QUE -1.59 -2.87 -0.31 0.0147 12 Potkin 2006 QUE 0.20 -0.08 0.48 0.1649 227 Small 1997 QUE -0.46 -0.83 -0.08 0.0164 111 Quetiapine pooled -0.35 -0.73 0.02 0.0658 652 Borison 1992 RIS -0.45 -0.83 -0.07 0.0208 107 Chouinard 1993 RIS -0.83 -1.44 -0.23 0.0070 44 Marder 1994 RIS -0.54 -0.90 -0.19 0.0025 127 Potkin 2003 RIS -0.48 -0.76 -0.20 0.0009 198 Potkin 2006 RIS -0.10 -0.38 0.18 0.4844 223 Study Ris-USA-72 1996 RIS -0.45 -0.73 -0.18 0.0011 236 Risperidone pooled -0.42 -0.59 -0.26 0.0000 935 van Kammen 1996 SER -0.26 -0.66 0.13 0.1898 98 Zborowski 1995 SER -0.20 -0.44 0.03 0.0832 325 Zimbroff 1997 SER -0.40 -0.70 -0.11 0.0077 191 Sertindole pooled -0.28 -0.44 -0.11 0.0011 614 Arato 2002 ZIP -0.59 -0.92 -0.26 0.0005 146 Daniel 1999 ZIP -0.45 -0.73 -0.16 0.0020 194 Keck 1998 ZIP -0.28 -0.70 0.14 0.1897 88 Study 115 2000 ZIP -0.30 -0.61 0.01 0.0612 156 Ziprasidone pooled -0.42 -0.58 -0.25 0.0000 584 Cooper 2000a ZOT -0.50 -0.88 -0.12 0.0106 106 Cooper 2000b ZOT -0.12 -0.48 0.24 0.5101 119 Moller 2004 ZOT -0.15 -0.59 0.29 0.4948 79 Zotepine pooled -0.26 -0.50 -0.02 0.0366 304 -2.00 -1.00 0.00 1.00 2.00 Favours Antipsychotic Favours Placebo Figure 7 Antipsychotic drugs versus placebo—negative symptoms. AMI, amisulpride; ARI, aripiprazole; CLO, clozapine; HAL, haloperidol; OLA, olanzapine; QUE, quetiapine; RIS, risperidone; SER, sertindole; ZIP, ziprasidone; ZOT, zotepine; circles are effect sizes of single studies, diamonds of pooled results.

Quality of life of use of antiparkinson medication or in terms of Two olanzapine studies20,41 found olanzapine signi- EPS rating scales. Only risperidone showed a trend ficantly superior to placebo on overall quality of of higher EPS on the rating scales (P = 0.075) (see life (N =2, n = 406, SMD = À0.38, CI: À0.59 to À0.17, Figures 9 and 10 and Table 2). P = 0.0003). Mo¨ller et al.23 found no significant superiority of zotepine (combined effect on the physical Sedation and the psychic components of the SF-36 scale: Although all antipsychotics were numerically more n = 72, SMD = À0.24, CI: À0.70 to 0.22, P = 0.309). sedating than placebo (see Figure 10), statistical No data on the other SGA drugs and haloperidol significance was reached only for haloperidol, queti- compared with placebo are available. apine and zotepine using the RR and for aripiprazole, haloperidol and quetiapine using the RD. The pooled Extrapyramidal adverse effects effect across SGA drugs was: N = 21, n = 3367, RR = In contrast to haloperidol, no SGA induced signi- 1.91, CI: 1.44–2.52; RD = 0.08, CI: 0.04–0.11, NNT = ficantly more EPS than placebo, whether in terms 13, CI: 9–25. It is not possible to disentangle the

Molecular Psychiatry Atypical antipsychotics versus placebo S Leucht et al 439 Study name Comparison Statistics for each study Hedges's g and 95% CI Hedges's Lower Upper p-Value Total g limit limit Danion 1999 AMI -0.49 -0.75 -0.22 0.0004 241 Palliere-Martinot 1995 AMI -0.60 -1.46 0.26 0.1705 20 Amisulpride pooled-0.50 -0.75 -0.24 0.0001 261 Honigfeld 1984a CLO -0.39 -1.24 0.45 0.3613 22 Clozapine pooled -0.39 -1.24 0.45 0.3613 22 Beasley 1996 HAL -0.35 -0.68 -0.01 0.0423 137 Study 115 2000 HAL -0.32 -0.63 -0.02 0.0397 162 Haloperidol pooled-0.33 -0.56 -0.11 0.0039 299 Beasley 1996 OLA -0.42 -0.71 -0.12 0.0055 201 Beasley 1996a OLA -0.23 -0.62 0.16 0.2423 100 Corrigan 2004 OLA -0.18 -0.47 0.11 0.2290 178 Olanzapine pooled-0.28 -0.47 -0.10 0.0024 479 Borison 1996 QUE 0.00 -0.38 0.38 1.0000 106 Potkin 2006 QUE -0.13 -0.41 0.15 0.3798 227 Quetiapine pooled-0.08 -0.31 0.14 0.4803 333 Potkin 2006 RIS -0.25 -0.54 0.03 0.0776 223 Study Ris-USA-72 1996 RIS 0.13 -0.14 0.40 0.3371 236 Risperidone pooled-0.06 -0.44 0.32 0.7614 459 van Kammen 1996 SER -0.30 -0.69 0.10 0.1413 98 Sertindole pooled -0.30 -0.69 0.10 0.1413 98 Study 115 2000 ZIP -0.42 -0.74 -0.11 0.0089 156 Daniel 1999 ZIP -0.19 -0.48 0.09 0.1883 189 Keck 1998 ZIP -0.51 -1.03 0.00 0.0517 59 Ziprasidone pooled-0.33 -0.52 -0.13 0.0011 404 Moller 2004 ZOT -0.48 -0.92 -0.03 0.0349 79 Zotepine pooled -0.48 -0.92 -0.03 0.0349 79 -2.00 -1.00 0.00 1.00 2.00 Favours Antipsychotic Favours Placebo Figure 8 Antipsychotic drugs versus placebo—depressive symptoms. AMI, amisulpride; ARI, aripiprazole; CLO, clozapine; HAL, haloperidol; OLA, olanzapine; QUE, quetiapine; RIS, risperidone; SER, sertindole; ZIP, ziprasidone; ZOT, zotepine; circles are effect sizes of single studies, diamonds of pooled results.

effects of concomitant benzodiazepines from those of P = 0.81, RD = 0.01, CI: À0.01 to 0.03, P = 0.46. In the the antipsychotics using meta-analysis (see Figure 11 sensitivity analysis, aripiprazole was even asso- and Table 2). ciated with fewer dropouts due to adverse events than placebo, while an increased risk was found for Dropout rates ziprasidone, haloperidol and sertindole. Amisulpride, olanzapine, sertindole and zotepine were not associated with significantly lower rates of Discussion all-cause dropouts than placebo, whereas aripiprazole, clozapine, quetiapine, risperidone and ziprasi- This review, based on 38 randomized controlled trials done were. This composite measure of efficacy, with 7323 participants, demonstrates the efficacy of tolerability and other factors has been used as a proxy SGA drugs over placebo on various measures of measure for acceptability of treatment.3 The overall response, relapse and discontinuation due to poor dropout rate when all studies were combined was as efficacy. Nevertheless, the relatively small absolute high as 47% (pooled ES SGA drugs versus PBO: difference in responder rates of 18%, translating into N = 37, n = 6001, RR = 0.75, CI: 0.69–0.82; RD = À0.14, an NNT of six, and the medium ES for the primary CI: À0.10 to À0.18, NNT = 7, CI: 6–10, P < 0.0001) outcome (change of overall symptoms) of À0.51 are (see Table 3). striking. Furthermore, we found that the drug– Dropouts due to insufficient efficacy confirmed that placebo difference diminished over time. This effect all antipsychotics were superior to placebo (pooled had already been reported in an analysis of psychia- results across SGA drugs: N = 36, n = 5809, RR = 0.52, tric trials by Trikalinos et al.57 CI: 0.45–0.59, RD = À0.17, CI: À0.20 to À0.13, Cohen58 described an ES of À0.50 as large enough to NNT = 6, CI: 5–8, P < 0.0001). be visible to the naked eye, for example, the No antipsychotic, not even haloperidol, was asso- difference between 14-year-old and 18-year-old girls ciated with significantly increased RR in terms of (about 1 inch) or the difference in IQ between clerical dropouts due to adverse events, but there was a and semiskilled workers. We pooled the (usually significantly increased RD for haloperidol and sertin- earlier) studies using the BPRS and found an absolute dole. The pooled ES across SGA drugs was also not difference of nine BPRS points between SGA drugs significant: N = 31, n = 5320, RR = 1.1, CI: 0.72–1.51, and placebo, which we translate into a difference of

Molecular Psychiatry Atypical antipsychotics versus placebo S Leucht et al 440 Study name Comparison Statistics for each study Risk ratio and 95% CI Risk Lower Upper p-Value Total ratio limit limit Boyer 1995 AMI 0.49 0.03 7.53 0.6056 104 Danion 1999 AMI 0.26 0.02 2.84 0.2698 242 Loo 1997 AMI 2.09 0.39 11.03 0.3865 141 Amisulpride pooled 0.87 0.24 3.20 0.8330 487 Kane 2002 ARI 0.92 0.49 1.74 0.7961 310 Modell 2005 ARI 1.25 0.45 3.45 0.6694 182 Pigott 2003 ARI 1.16 0.65 2.05 0.6155 310 Study 138001 2002 ARI 1.31 0.60 2.88 0.5000 314 Study 93202 2002 ARI 0.62 0.25 1.51 0.2915 69 Study 94202 2002 ARI 1.16 0.69 1.93 0.5708 125 Aripiprazole pooled 1.07 0.81 1.41 0.6432 1310 Arvanitis 1997 HAL 3.50 1.66 7.37 0.0010 103 Beasley 1996 HAL 5.79 2.96 11.32 0.0000 137 Borison 1992 HAL 2.11 1.26 3.53 0.0043 107 Chouinard 1993 HAL 2.62 1.26 5.46 0.0101 43 Kane 2002 HAL 2.43 1.35 4.38 0.0031 210 Marder 1994 HAL 2.58 1.46 4.58 0.0012 132 Study 115 2000 HAL 1.61 1.10 2.37 0.0138 168 Study 93202 2002 HAL 1.96 1.07 3.57 0.0292 69 Study 94202 2002 HAL 1.50 0.94 2.39 0.0906 127 Zborowski 1995 HAL 2.65 1.67 4.21 0.0000 231 Zimbroff 1997 HAL 2.41 1.47 3.94 0.0005 281 Haloperidol pooled 2.34 1.90 2.88 0.0000 1608 Beasley 1996 OLA 2.11 1.03 4.31 0.0408 201 Beasley 1996a OLA 0.50 0.16 1.55 0.2311 100 Corrigan 2004 OLA 1.40 0.41 4.81 0.5896 180 Olanzapine pooled 1.23 0.52 2.93 0.6403 481 Arvanitis 1997 QUE 0.84 0.37 1.88 0.6647 208 Borison 1996 QUE 0.85 0.28 2.62 0.7753 109 Small 1997 QUE 0.70 0.28 1.76 0.4491 192 Quetiapine pooled 0.79 0.46 1.35 0.3888 509 Bai 2003 RIS 1.27 0.74 2.18 0.3816 42 Borison 1992 RIS 1.31 0.73 2.36 0.3670 107 Chouinard 1993 RIS 1.17 0.47 2.92 0.7416 44 Marder 1994 RIS 1.12 0.55 2.26 0.7581 130 Risperidone pooled 1.24 0.89 1.71 0.2004 323 van Kammen 1996 SER 0.10 0.01 1.79 0.1176 102 Zborowski 1995 SER 0.90 0.54 1.51 0.6963 346 Zimbroff 1997 SER 0.74 0.40 1.39 0.3531 213 Sertindole pooled 0.79 0.51 1.23 0.2998 661 Arato 2002 ZIP 1.48 0.70 3.11 0.3025 146 Daniel 1999 ZIP 1.92 1.03 3.58 0.0410 196 Keck 1998 ZIP 2.30 0.76 6.95 0.1407 95 Study 115 2000 ZIP 0.61 0.35 1.07 0.0850 161 Ziprasidone pooled 1.33 0.70 2.51 0.3856 598 Cooper 2000a ZOT 1.33 0.31 5.67 0.6969 106 Cooper 2000b ZOT 1.61 0.50 5.22 0.4265 121 Zotepine pooled 1.49 0.60 3.72 0.3883 227 0.01 0.1 1 10 100 Favours Antipsychotic Favours Placebo Figure 9 Antipsychotic drugs versus placebo—use of antiparkinson medication. AMI, amisulpride; ARI, aripiprazole; CLO, clozapine; HAL, haloperidol; OLA, olanzapine; QUE, quetiapine; RIS, risperidone; SER, sertindole; ZIP, ziprasidone; ZOT, zotepine, circles are effect sizes of single studies, diamonds of pooled results.

one point on the Clinical Global Impression Scale.59 more chronic participants in our studies, half of We pooled the more recent studies using the PANSS the participants had a first episode of schizophrenia and found a difference of 10 points. According to and received antipsychotic drugs for the first time. Leucht et al.,59 a PANSS total score difference of The Cochrane Review comparing the standard 15 points reflects minimal improvement according to drug chlorpromazine to placebo found an NNT of the CGI. 4 in the short-term (n = 590, 11 studies, response rate The meta-analysis confirms that the SGA drugs are drug 65.9%, placebo 41.5%, weighted RD 25%) and no ‘wonder drugs’ in terms of efficacy, and that there 6 in the medium term (n = 1121, 13 randomized is much room for better medication, confirming recent controlled trials, response rate drug 28.1%, response naturalistic studies.60,61 But the fact that the ESs of the rate placebo 13.1%, weighted RD 18%).63 The haloperidol arm studies also revealed a moderate Cochrane Review on haloperidol showed a pronoun- effect raises the question whether antipsychotic drugs ced superiority over placebo, with an NNT of 3 in have previously been overestimated. An early NIMH both short- and medium-term studies (medium-term study has often been quoted as a proof for a strong results: n = 308, eight studies, response rate drug effect of antipsychotic drugs.62 In this study (n = 344), 43.8%, placebo 14.4%, weighted RD 32%64). An older the response rate to drug was 61% compared with review on maintenance treatment found substantially 22% in the placebo group, resulting in a response rate lower relapse rates of 16% in the antipsychotic group difference of 41% or an NNT of 2. In contrast to the compared with 53% in the placebo group.65 The

Molecular Psychiatry Atypical antipsychotics versus placebo S Leucht et al 441 Study name Comparison Statistics for each study Hedges's g and 95% CI Hedges's Lower Upper p-Value Total g limit limit Kane 2002 ARI 0.11 -0.13 0.34 0.3768 310 Modell 2005 ARI 0.21 -0.09 0.51 0.1699 169 Pigott 2003 ARI -0.24 -0.46 -0.01 0.0394 306 Potkin 2003 ARI 0.04 -0.20 0.27 0.7561 304 Study 138001 2002 ARI 0.00 -0.23 0.23 1.0000 305 Study 94202 2002 ARI -0.14 -0.49 0.21 0.4307 125 Aripiprazole pooled -0.01 -0.14 0.12 0.9191 1519 Arvanitis 1997 HAL 0.37 -0.03 0.76 0.0674 99 Beasley 1996 HAL 0.50 0.15 0.85 0.0046 131 Chouinard 1993 HAL 0.70 0.10 1.31 0.0229 43 Kane 2002 HAL 0.38 0.10 0.65 0.0066 212 Marder 1994 HAL 0.41 0.07 0.76 0.0200 127 Study 94202 2002 HAL 0.18 -0.17 0.52 0.3210 127 Zimbroff 1997 HAL 0.40 0.13 0.68 0.0041 265 Haloperidol pooled 0.39 0.26 0.52 0.0000 1004 Beasley 1996 OLA 0.14 -0.16 0.44 0.3713 187 Beasley 1996a OLA 0.04 -0.35 0.43 0.8300 99 Beasley 2003 OLA -0.03 -0.26 0.20 0.7979 326 Risperidone pooled 0.03 -0.13 0.20 0.6861 612 Arvanitis 1997 QUE -0.33 -0.65 -0.01 0.0424 197 Borison 1996 QUE 0.08 -0.31 0.46 0.6964 103 Potkin 2006 QUE 0.00 -0.28 0.28 1.0000 229 Quetiapine pooled -0.09 -0.33 0.15 0.4575 529 Bai 2003 RIS 0.07 -0.53 0.66 0.8292 42 Chouinard 1993 RIS -0.02 -0.60 0.56 0.9349 44 Marder 1994 RIS 0.09 -0.26 0.43 0.6231 128 Potkin 2003 RIS 0.03 -0.25 0.30 0.8541 202 Potkin 2006 RIS 0.36 0.08 0.64 0.0119 226 Sertindole pooled 0.14 -0.01 0.30 0.0748 642 van Kammen 1996 SER 0.17 -0.22 0.56 0.3976 100 Zimbroff 1997 SER 0.02 -0.27 0.31 0.9103 199 Ziprasidone pooled 0.07 -0.16 0.30 0.5512 299 Keck 1998 ZIP 0.12 -0.28 0.52 0.5668 95 Zotepine pooled 0.12 -0.28 0.52 0.5668 95 Cooper 2000a ZOT 0.14 -0.24 0.52 0.4679 106 Moller 2004 ZOT -0.15 -0.59 0.29 0.4990 79 Olanzapine pooled 0.02 -0.27 0.30 0.9149 185 -1.00 -0.50 0.00 0.50 1.00 Favours Antipsychotic Favours Placebo Figure 10 Antipsychotics versus placebo—extrapyramidal symptoms rating scales. AMI, amisulpride; ARI, aripiprazole; CLO, clozapine; HAL, haloperidol; OLA, olanzapine; QUE, quetiapine; RIS, risperidone; SER, sertindole; ZIP, ziprasidone; ZOT, zotepine; circles are effect sizes of single studies, diamonds of pooled results. relapse results of our review also suggested a more the eligible schizophrenic patients are entered into pronounced long-term superiority of SGA drugs, at clinical trials.68,69 ‘Failed studies’ in which neither least in terms of RRs. In summary, these reviews haloperidol nor the SGA drugs were better than highlight that there is a substantial placebo response, placebo cannot explain the relatively small differ- which in our sample was 24% based on a response ence, because the pooled ES of studies with a definition of at least 20–30% total score reduction in significantly effective haloperidol arm was similar two-thirds of the studies. Consistent with our meta- (Hedges’s g = À0.54, RD = À0.16). The high dropout regression analyzing publication year as a moderator, rates in the studies (overall 47%) may decrease the the degree of improvement seems to decrease over drug–placebo difference, because the antipsychotic time. drugs do not have time to develop their full An obvious question is whether design issues can, effects, and the full deterioration under placebo is at least partly, account for these findings. A mean age also decreased if participants are prematurely taken of 37.5 years suggests that the participants were out of the trial. The Cochrane Review on haloperidol relatively chronic. Less chronic patients respond excluded studies with dropout rates higher than better to antipsychotic drugs. For example, the mean 50% and found an NNT of 3.64 We did not apply age in the early NIMH study mentioned above was such an approach, because it is not clear what degree 28.2 years.62 Remission rates of more than 80% have of attrition will clearly bias the results and in been achieved in 1-year studies of first-episode which direction. On the other hand, the studies in patients.66,67 The generalizability of recent studies is the Cochrane Review were all published before 1993 called into question by the fact that only 10–15% of and were arguably less ‘sophisticated,’ for example,

Molecular Psychiatry Atypical antipsychotics versus placebo S Leucht et al 442 Study name Comparison Statistics for each study Risk ratio and 95% CI Risk Lower Upper p-Value Total ratio limit limit Kane 2002 ARI 1.95 0.66 5.72 0.2251 310 Modell 2005 ARI 6.56 0.34 125.18 0.2113 182 Pigott 2003 ARI 1.67 0.41 6.85 0.4789 310 Potkin 2003 ARI 1.07 0.54 2.10 0.8531 305 Aripiprazole pooled 1.38 0.82 2.34 0.2240 1107 Arvanitis 1997 HAL 3.43 1.21 9.73 0.0203 103 Beasley 1996 HAL 2.15 1.14 4.04 0.0173 137 Borison 1992 HAL 7.13 0.91 56.00 0.0617 107 Kane 2002 HAL 3.31 1.12 9.83 0.0309 210 Marder 1994 HAL 7.00 0.37 132.91 0.1951 132 Zimbroff 1997 HAL 0.75 0.42 1.34 0.3320 281 Haloperidol pooled 2.28 1.11 4.67 0.0241 970 Beasley 1996 OLA 2.14 1.19 3.85 0.0115 201 Beasley 1996a OLA 0.86 0.31 2.37 0.7665 100 Kryzhanovskaya 2006 OLA 8.26 1.15 59.61 0.0362 107 Risperidone pooled 1.93 0.76 4.90 0.1655 408 Arvanitis 1997 QUE 1.14 0.39 3.30 0.8133 208 Borison 1996 QUE 5.35 1.96 14.55 0.0010 109 Fabre 1995 QUE 5.00 0.33 75.11 0.2443 12 Potkin 2006 QUE 1.50 0.57 3.93 0.4122 229 Small 1997 QUE 1.71 0.95 3.11 0.0760 192 Quetiapine pooled 2.02 1.18 3.47 0.0107 750 Borison 1992 RIS 2.04 0.19 21.81 0.5561 107 Marder 1994 RIS 5.15 0.25 105.31 0.2868 130 Potkin 2003 RIS 1.32 0.63 2.78 0.4571 202 Potkin 2006 RIS 0.95 0.34 2.69 0.9294 226 Sertindole pooled 1.29 0.73 2.29 0.3834 665 van Kammen 1996 SER 2.67 0.56 12.59 0.2156 102 Zimbroff 1997 SER 0.97 0.54 1.74 0.9142 213 Ziprasidone pooled 1.23 0.53 2.87 0.6285 315 Daniel 1999 ZIP 3.54 1.38 9.05 0.0083 196 Keck 1998 ZIP 1.02 0.27 3.85 0.9752 95 Zotepine pooled 2.08 0.62 6.95 0.2321 291 Cooper 2000a ZOT 35.00 2.16 567.39 0.0124 106 Cooper 2000b ZOT 2.85 1.54 5.29 0.0009 121 Moller 2004 ZOT 3.52 0.15 84.15 0.4364 85 Olanzapine pooled 4.60 1.21 17.50 0.0252 312 0.01 0.1 1 10 100 Favours Antipsychotics Favours Placebo Figure 11 Antipsychotics versus placebo—sedation. AMI, amisulpride; ARI, aripiprazole; CLO, clozapine; HAL, haloperidol; OLA, olanzapine; QUE, quetiapine; RIS, risperidone; SER, sertindole; ZIP, ziprasidone; ZOT, zotepine, circles are effect sizes of single studies; diamonds of pooled results.

because some did not use standardized diagnostic practice. On the one hand, the data clearly show that criteria and rating scales, had small sample sizes or the SGA drugs are no wonder drugs in terms of were carried out in single centers. It should also be efficacy, producing a moderate ES (0.49) in compar- noted that haloperidol is a high-potency antipsycho- ison with placebo. But what then do the ESs of those tic drug. CATIE and CUtLASS suggested that these SGA drugs that were more efficacious than first- results are not necessarily representative for low- generation antipsychotics in the analysis by Davis potency or intermediate-potency antipsychotics.60,61 et al.4 mean? They ranged between 0.21 (olanzapine) Preliminary work by Davis et al.4 had revealed a and 0.49 (clozapine). similar ES for the difference between haloperidol and Other problems are evident when negative symp- placebo equaling 0.60. toms and depression are considered. With the The funnel plots raised the possibility of publi- exception of amisulpride and olanzapine (5 but not cation bias. This method must be interpreted 20 mg dayÀ120), there is still no proof that SGA drugs with caution, because there can be other reasons for are effective for predominantly negative symptoms, the plot asymmetry, especially true heterogeneity, because populations with predominantly positive because studies with different SGA drugs and possi- symptoms are simply not appropriate to examine this bly different efficacy were pooled.30 Another issue is issue due to secondary effects, and statistical methods that almost all included studies were organized by such as path analysis can only, in part, account pharmaceutical companies, who may have not pub- for these effects.71 Even more surprising was that lished studies with small drug–placebo difference, haloperidol decreases not only negative symptoms, raising the possibility of an ‘industry bias.’70 Never- but also depressive symptoms significantly more theless, even if methodological issues accounted, than placebo. It has been said that conventional in part, for the small differences, it is difficult to antipsychotics induce depression rather than allevi- interpret the effectiveness of SGA drugs in clinical ate it.72 It may be that the depression-inducing effect

Molecular Psychiatry Table 2 Risk differences and numbers needed to treat/harm for the outcomes non-response, relapse, antiparkinson medication and somnolence

Non-response

Antipsychotic drug Nn Antipsychotic non-response, % Placebo non-response, % Risk difference (CI) P-value NNT (CI)

Amisulpride 3 487 50.0 76.7 À0.25 (À0.35 to À0.15) 0 4 (3–7) Aripiprazole 5 1123 63.8 78.6 À0.15 (À0.20 to À0.10) 0 7 (5–10) Haloperidol 10 1440 70.7 80.5 À0.12 (À0.17 to À0.07) 0 9 (6–15) Olanzapine 4 582 53.4 73.3 À0.14 (À0.22 to À0.06) 0.0005 7 (5–16) Quetiapine 5 750 56.9 65.9 À0.09 (À0.19 to 0.01) 0.0748 NE Risperidone 7 997 43.6 69.6 À0.28 (À0.39 to À0.18) 0 4 (3–6) Sertindole 3 661 82.5 89.0 À0.08 (À0.17 to 0.02) 0.1077 NE Ziprasidone 3 291 65.6 80.0 À0.15 (À0.25 to À0.05) 0.0042 7 (4–22) Zotepine 2 227 56.0 75.7 À0.21 (À0.45 to 0.04) 0.1036 NE Relapse

Antipsychotic drug Nn Antipsychotic relapsed, % Placebo relapsed, % Risk difference P-value NNT (CI)

Amisulpride 2 246 7.9 13.1 À0.07 (À0.21 to 0.07) 0.3439 NE Aripiprazole 1 310 32.3 54.8 À0.23 (À0.33 to À0.12) 0 4 (3–8) Olanzapine 3 559 8.3 29.3 À0.22 (À0.29 to À0.14) 0 5 (3–7) Ziprasidone 1 277 34.5 60.6 À0.26 (À0.39 to À0.13) 0.0001 4 (3–8) Zotepine 2 119 6.6 36.2 À0.30 (À0.43 to À0.16) 0 3 (2–6) Use of antiparkinson medication at least once

Antipsychotic drug NnAntipsychotic antiparkinson medication, % Placebo antiparkinson medication, % Risk difference P-value NNH (CI) Leucht S placebo versus antipsychotics Atypical

Amisulpride 4 514 1.9 2.5 À0.01 (À0.04 to 0.02) 0.5406 NE Aripiprazole 6 1310 13.3 13.5 0.01 (À0.02 to 0.05) 0.5328 NE al et Haloperidol 11 1608 47.6 20.1 0.29 (0.22 to 0.36) 0 3 (3–5) Olanzapine 3 481 15.6 9.8 0.03 (À0.07 to 0.13) 0.6054 NE Quetiapine 4 521 9.5 11.2 À0.02 (À0.08 to 0.03) 0.3821 NE Risperidone 4 323 32.3 25.9 0.05 (À0.04 to 0.15) 0.2673 NE Sertindole 3 661 12.7 15.6 À0.05 (À0.10 to 0.00) 0.0650 NE Ziprasidone 4 598 21.3 17.4 0.05 (À0.06 to 0.15) 0.4001 NE Zotepine 2 227 9.5 6.3 0.03 (À0.04 to 0.10) 0.3998 NE Somnolence/sedation

Antipsychotic drug Nn Antipsychotic somnolence, % Placebo somnolence, % Risk difference P-value NNH (CI)

Aripiprazole 4 1107 7.0 4.0 0.02 (0.00 to 0.05) 0.0494 44 (22–16 838) Haloperidol 6 970 16.5 8.1 0.08 (0.02 to 0.14) 0.0069 12 (7–44) Olanzapine 3 408 27.1 12.4 0.13 (À0.01 to 0.26) 0.0661 NE Quetiapine 5 750 16.8 9.7 0.13 (0.01 to 0.24) 0.0283 8 (4–74) Risperidone 4 665 7.6 5.7 0.02 (À0.01 to 0.05) 0.1973 NE Sertindole 2 315 16.5 13.2 0.04 (À0.04 to 0.11) 0.3561 NE Ziprasidone 2 291 15.9 6.4 0.07 (À0.06 to 0.21) 0.2745 NE

oeua Psychiatry Molecular Zotepine 3 312 31.6 6.4 0.21 (À0.21 to 0.44) 0.0619 NE

Abbreviations: CI, 95% confidence interval; N, number of studies; n, number of patients; NE, not estimable, because the risk difference did not show a significant difference between groups; NNH, number needed to harm. 443 Atypical antipsychotics versus placebo S Leucht et al 444 Table 3 Summary results of the outcomes dropout rates (any reason, inefficacy of treatment and adverse events)

Dropout due to any reason

Antipsychotic NnAntipsychotic, Placebo, Relative risk P-value Risk difference P-value NNT (CI) % % (CI)

Amisulpride 5 618 29.1 47.0 0.69 (0.48–1.00) 0.0523 À0.13 (À0.25 to À0.01) 0.0274 8 (4–68) Aripiprazole 7 1615 46.5 59.0 0.80 (0.72–0.89) 0 À0.12 (À0.17 to À0.07) 0 8 (6–14) Clozapine 1 24 37.5 100.0 0.40 (0.22–0.76) 0.0045 À0.62 (À0.90 to À0.35) 0 2 (1–3) Haloperidol 11 1608 50.8 61.0 0.84 (0.77–0.92) 0.0001 À0.10 (À0.15 to À0.05) 0.0001 10 (7–20) Olanzapine 6 1088 36.0 52.4 0.70 (0.46–1.05) 0.0815 À0.14 (À0.31 to 0.02) 0.0887 NE Quetiapine 5 750 37.8 49.8 0.79 (0.68–0.92) 0.0027 À0.09 (À0.16 to À0.02) 0.0095 11 (6–46) Risperidone 6 955 29.4 47.1 0.70 (0.57–0.86) 0.0009 À0.16 (À0.26 to À0.05) 0.0028 6 (4–18) Sertindole 3 661 54.2 55.7 0.96 (0.83–1.10) 0.5509 À0.02 (À0.10 to 0.06) 0.5886 NE Ziprasidone 4 598 45.7 63.1 0.73 (0.63–0.84) 0 À0.17 (À0.26 to À0.07) 0.0006 6 (4–14) Zotepine 3 312 39.4 39.5 0.94 (0.64–1.38) 0.7709 À0.03 (À0.15 to 0.10) 0.6875 NE Dropout due to inefficacy of treatment

Antipsychotic NnAntipsychotic, Placebo, Risk ratio P-value Risk difference P-value NNT (CI) % %

Amisulpride 4 514 9.9 28.2 0.48 (0.33–0.69) 0.0001 À0.12 (À0.18 to À0.06) 0.0001 8 (6–17) Aripiprazole 7 1615 13.2 25.2 0.61 (0.48–0.77) 0 À0.09 (À0.16 to À0.02) 0.0085 11 (6–43) Clozapine 1 24 0 62.5 0.05 (0.00–0.78) 0.0324 À0.62 (À0.95 to À0.30) 0.0002 2 (1–3) Haloperidol 11 1608 20.7 39.4 0.55 (0.46–0.65) 0 À0.18 (À0.23 to À0.13) 0 6 (4–8) Olanzapine 6 1088 15.3 36.7 0.42 (0.25–0.69) 0.0007 À0.18 (À0.26 to À0.10) 0 6 (4–10) Quetiapine 4 521 31.1 48.5 0.60 (0.49–0.75) 0 À0.20 (À0.28 to À0.11) 0 5 (4–9) Risperidone 5 729 9.0 30.2 0.34 (0.24–0.49) 0 À0.23 (À0.37 to À0.09) 0.0015 4 (3–11) Sertindole 3 661 25.9 37.6 0.68 (0.54–0.86) 0.0012 À0.12 (À0.20 to À0.05) 0.0015 8 (5–22) Ziprasidone 4 598 23.3 40.9 0.59 (0.46–0.74) 0 À0.17 (À0.24 to À0.10) 0 6 (4–10) Zotepine 3 312 9.0 30.6 0.29 (0.17–0.49) 0 À0.20 (À0.41 to 0.01) 0.0601 NE

Dropout due to adverse events

Antipsychotic NnAntipsychotic, Placebo, Risk ratio P-value Risk difference P-value NNH (CI) % %

Amisulpride 4 514 5.8 9.4 0.59 (0.20–1.78) 0.3474 À0.02 (À0.08 to 0.03) 0.3620 NE Aripiprazole 7 1615 7.1 9.1 0.73 (0.49–1.08) 0.1194 À0.01 (À0.04 to 0.02) 0.3952 NE Clozapine 1 24 6.3 12.5 0.50 (0.04–7.00) 0.6067 À0.06 (À0.32 to 0.20) 0.6350 NE Haloperidol 10 1476 8.7 6.4 1.31 (0.80–2.12) 0.2817 0.04 (0.02 to 0.07) 0.0016 25 (15–65) Olanzapine 5 914 3.7 6.4 0.83 (0.18–3.84) 0.8069 0.00 (À0.07 to 0.06) 0.8976 NE Quetiapine 4 521 3.5 3.4 1.10 (0.27–4.38) 0.8974 0.00 (À0.03 to 0.04) 0.8248 NE Risperidone 4 599 7.7 9.5 0.86 (0.35–2.13) 0.7466 0.00 (À0.08 to 0.08) 0.9617 NE Sertindole 3 661 8.7 4.6 1.86 (0.96–3.60) 0.0643 0.04 (0.01 to 0.08) 0.0131 22 (13–107) Ziprasidone 4 598 5.0 2.3 2.49 (0.33–18.89) 0.3773 0.03 (À0.03 to 0.08) 0.3730 NE Zotepine 3 312 12.3 5.1 1.88 (0.61–5.85) 0.2742 0.05 (À0.06 to 0.15) 0.3917 NE

is a long-term one, while haloperidol improves that some SGA drugs do induce EPS.73 There is depression in the short run. But it is also possible some evidence that bipolar patients are more sensi- that all these symptoms are truly related and the tive to EPS than people with schizophrenia.74 But expression of the same underlying pathology. many participants in schizophrenia trials were pre- While there was substantial evidence that haloper- viously treated with antipsychotics for long periods idol produces EPS, none of the SGA drugs induced with washout periods often lasting only a few days. significantly more EPS than placebo. This finding In contrast, many mania patients had much less demonstrates that the EPS risks of all SGA drugs are exposure to antipsychotics. Consequently, carry- low, but it does not prove that they are all free from over effects of prior treatment may have reduced these adverse effects. A meta-analysis comparing drug–placebo differences. Indeed, if overall rates of SGA drugs with placebo in bipolar mania suggested use of antiparkinson medication under the different

Molecular Psychiatry Atypical antipsychotics versus placebo S Leucht et al 445 SGA drugs are considered rather than ESs compared 2 Moncrieff J, Kirsch I. Efficacy of antidepressants in adults. with placebo, these rates were considerable for some BMJ 2005; 331: 155–157. SGA drugs (amisulpride 2%—please note that these 3 Adams CE, Coutinho E, Davis JH, Duggan L, Leucht S, Tharyan P, À1 Cochrane Schizophrenia Group In: The Cochrane Library. John were low doses up to only 300 mg day , aripiprazole Wiley & Sons Ltd: Chichester, UK, 2007. 13.3%, clozapine—no data available, olanzapine 4 Davis JM, Chen N, Glick ID. A meta-analysis of the efficacy of 15.6%, quetiapine 9.5%, risperidone 32.3%, sertin- second-generation antipsychotics. Arch Gen Psychiatry 2003; 60: dole 12.7%, ziprasidone 21.3%, zotepine 9.5%), 553–564. although all were clearly lower than for haloperidol 5 Geddes J, Freemantle N, Harrison P, Bebbington P. Atypical antipsychotics in the treatment of schizophrenia: systematic (47.6%). The placebo rates varied from 2.5% (ami- overview and meta-regression analysis. BMJ 2000; 321: 1371–1376. sulpride studies) to 25.9% (risperidone studies). 6 Higgins JPT, Green S. Cochrane Handbook for Systematic Reviews Longer washout periods would improve the sensiti- of Interventions 4.2.5. In: The Cochrane Library. John Wiley & Sons vity for detecting EPS differences but are problematic Ltd: Chichester, UK, 2005. from an ethical point of view. 7 Arvanitis LA, Miller BG, Seroquel Trial 13 Study Group. Multiple fixed doses of ‘seroquel’ (quetiapine) in patients with acute There were few differences between antipsychotics exacerbation of schizophrenia: a comparison with haloperidol and placebo in terms of dropouts due to adverse and placebo. Biol Psychiatry 1997; 42: 233–246. events. Unfortunately, some efficacy-related adverse 8 Kay SR, Fiszbein A, Opler LA. The positive and negative events such as agitation due to insufficient efficacy syndrome scale (PANSS) for schizophrenia. Schizophr Bull 1987; 13: 261–275. are counted as adverse events and falsely inflate 9 Overall JE, Gorham DR. The brief psychiatric rating scale. Psychol the placebo ‘adverse event’ tabulation. For example, Rep 1962; 10: 790–812. we found fewer dropouts due to adverse events with 10 Guy W. Clinical Global Impression. In: ECDEU Assessment aripiprazole than with placebo. We suggest that Manual for Psychopharmacology, Revised (DHEW Publ No ADM only adverse events reflecting side effects should be 76-338). National Institute of Mental Health: Rockville, MD, 1976, pp 218–222. presented as dropouts due to adverse events to make 11 Simpson M, Angus JW. A rating scale for extrapyramidal side this outcome a useful global measure of tolerability. effects. Acta Psychiatr Scand Suppl 1970; 212: 11–19. For perspective, even internal medicine drugs 12 Chouinard G, Ross-Chouinard A, Annable L, Jones BD. The rarely cure. It is impossible to make direct compari- extrapyramidal symptom rating scale. Can J Neurol Sci 1980; 7: 233. sons of different treatments of different diseases, 13 Smeeth L, Haines A, Ebrahim S. Numbers needed to treat derived but (to give an example of a contemporary drug) the from meta-analyses—sometimes informative, usually misleading. NNT to avoid vascular events or death by statins is BMJ 1999; 318: 1548–1551. higher than 100.75 The implication of our findings for 14 Elmore JG, Gigerenzer G. Benign breast disease—the risks of antipsychotic drug development is that there is a great communicating risk. N Engl J Med 2005; 353: 297–299. 15 Boissel JP, Cucherat M, Li W, Chatellier G, Gueyffier F, Buyse M deal of room for improvement, let alone cure. et al. The problem of therapeutic efficacy indices. 3. Comparison of the indices and their use. Therapie 1999; 54: 405–411. 16 Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring Acknowledgments inconsistency in meta-analyses. BMJ 2003; 327: 557–560. 17 Der-Simonian R, Laird N. Meta-analysis in clinical trials. Control We are much indebted to the Cochrane Schizophrenia Clin Trials 1986; 7: 177–188. Group. Without access to its register of randomized 18 Boyer P, Lecrubier Y, Puech AJ, Dewailly J, Aubin F. Treatment of negative symptoms in schizophrenia with amisulpride. Br J controlled trials, this review would not have been Psychiatry 1995; 166: 68–72. possible. We also thank AstraZeneca, BristolMyersS- 19 Danion JM, Rein W, Fleurot O. Improvement of schizophrenic quibb, EliLilly, Lundbeck and Sanofi-Aventis for patients with primary negative symptoms treated with amisul- providing unpublished data. pride. Am J Psychiatry 1999; 156: 610–616. 20 Lecrubier Y, Bouhassira M, Olivier V, Lancrenon S, Crawford AM. Olanzapine versus amisulpride and placebo in the treatment of negative symptoms and deficit states of chronic schizophrenia. Disclosure/Conflict of interest Eur Neuropsychopharmacol 1999; 9: S288. 21 Loo H, Poirier-Littre MF, Theron M, Rein W, Fleurot O. This meta-analysis received no funding. Stefan Amisulpride versus placebo in the medium-term treatment of Leucht has received speaker or consultancy honoraria the negative symptoms of schizophrenia. Br J Psychiatry 1997; from Sanofi-Aventis, BMS, Lilly, Janssen, Lundbeck 170: 18–22. and Pfizer. Lilly and Sanofi-Aventis sponsored 22 Paille`re Martinot ML, Lecrubier Y, Martinot JL, Aubin F. Improve- ment of some schizophrenic deficit symptoms with low doses of research projects by Dr Leucht. Werner Kissling has amisulpride. Am J Psychiatry 1995; 152: 130–134. received speaker or consultancy honoraria from 23 Mo¨ller HJ, Riedel M, Mu¨ller N, Fischer W, Kohnen R. Zotepine Sanofi-Aventis, BMS, Lilly, Janssen, Lundbeck, Bayer versus placebo in the treatment of schizophrenic patients with and Pfizer. Dieter Arbter, Rolf R Engel and John M stable primary negative symptoms: a randomized double-blind multicenter trial. Pharmacopsychiatry 2004; 37: 270–278. Davis have no conflict of interest to declare. 24 Pigott TA, Carson WH, Saha AR, Torbeyns AF, Stock EG, Ingenito GG. Aripiprazole for the prevention of relapse in stabilized patients with chronic schizophrenia: a placebo-controlled 26- References week study. J Clin Psychiatry 2003; 64: 1048–1056. 25 Bai YM, Yu SC, Lin CC. Risperidone for severe tardive dyskinesia: 1 Bentham P, Gray R, Raftery J, Hills R, Sellwood E, Courtney C et al. a 12-week randomized, double-blind, placebo-controlled study. Long-term donepezil treatment in 565 patients with Alzheimer’s J Clin Psychiatry 2003; 64: 1342–1348. disease (AD2000): randomised double-blind trial. Lancet 2004; 26 Arato M, O’Connor R, Meltzer H, Zeus Study Group. Ziprasidone 363: 2105–2115. in the long-term treatment of negative symptoms and the

Molecular Psychiatry Atypical antipsychotics versus placebo S Leucht et al 446 prevention of exacerbation of schizophrenia. Int Clin Psycho- treatment of chronic schizophrenic patients. J Clin Psycho- pharmacol 2002; 17: 207–215. pharmacol 1993; 13: 25–40. 27 Cooper SJ, Butler A, Tweed J, Welch C, Raniwalla J. Zotepine in 49 Marder SR, Meibach RC. Risperidone in the treatment of the prevention of recurrence: a randomised, double-blind, place- schizophrenia. Am J Psychiatry 1994; 151: 825–835. bo-controlled study for chronic schizophrenia. Psychopharmaco- 50 Office of Clinical Pharmacology and Biopharmacy Review. logy 2000a; 150: 237–243. Study Ris-USA-72. NDA number: 20272. Janssen-Cilag, Data on 28 Cooper SJ, Tweed J, Raniwalla J, Butler A, Welch C. A placebo File 1996. controlled comparison of zotepine versus chlorpromazine in 51 van Kammen DP. A randomized, controlled, dose-ranging trial of patients with acute exacerbation of schizophrenia. Acta Psychiatr sertindole in patients with schizophrenia. Psychopharmacology Scand 2000b; 101: 218–225. 1996; 124: 168–175. 29 Potkin SG, Gharabawi GM, Greenspan AJ, Mahmoud R, Kosik- 52 Zborowski J, Schmitz P, Staser J, O’Neil J, Giles K, Wallin B et al. Gonzalez C, Rupnow MF et al. A double-blind comparison of Efficacy and safety of sertindole in a trial of schizophrenic risperidone, quetiapine and placebo in patients with schizo- patients. Presented at the Congress of the American Psychiatric phrenia experiencing an acute exacerbation requiring hospitaliza- Association in Miami 1995. tion. Schizophr Res 2006; 85: 254–265. 53 Zimbroff DL, Kane JM, Tamminga CA, Daniel DG, Mack RJ, 30 Egger M, Davey SG, Schneider M, Minder C. Bias in meta-analysis Wozniak PJ et al. Controlled, dose response study of sertindole detected by a simple, graphical test. BMJ 1997; 315: 629–634. and haloperidol in the treatment of schizophrenia. Am J 31 Borenstein M, Hedges LV, Higgins JPT, Rothstein H. Compre- Psychiatry 1997; 154: 782–791. hensive Meta-analysis Version 2 http://www meta-analysis com, 54 Daniel DG, Zimbroff DL, Potkin SG, Reeves KR, Harrigan EP, 2006. Lakshminarayanan M. Ziprasidone 80 and 160 mg/day in the 32 Kane JM, Carson WH, Saha AR, McQuade RD, Ingenito GG, acute exacerbation of schizophrenia and schizoaffective disorder: Zimbroff DL et al. Efficacy and safety of aripiprazole and a 6-week placebo-controlled trial. Neuropsychopharmacology haloperidol versus placebo in patients with schizophrenia and 1999; 20: 491–505. schizoaffective disorder. J Clin Psychiatry 2002; 63: 763–771. 55 Keck P, Buffenstein A, Ferguson J, Feighner J, Jaffe W, Harrigan EP 33 Modell S, Werner C, Spevakne-Gorocs T, Kungel M, Ebrecht M. et al. Ziprasidone 40 and 120 mg/day in the acute exacerbation of Efficacy and safety of lower doses of aripiprazole. Pharmaco- schizophrenia and schizoaffective disorder: a 4-week placebo psychiatry 2005; 38: 264. controlled trial. Psychopharmacology 1998; 140: 173–184. 34 Potkin SG, Saha AR, Kujawa MJ, Carson WH, Ali M, Stock E et al. 56 Center for Drug Evaluation and Research. Study 115. Approval Aripiprazole, an antipsychotic with a novel mechanism of action, package for application number 20-825. Medical review. and risperidone vs placebo in patients with schizophrenia and http://www fda gov, 2000. schizoaffective disorder. Arch Gen Psychiatry 2003; 60: 681–690. 57 Trikalinos TA, Churchill R, Ferri M, Leucht S, Tuunainen A, 35 Center for Drug Evaluation and Research. Study 138001. Applica- Wahlbeck K et al. Effect sizes in cumulative meta-analyses tion number 21-436. Medical review(s). http://www.fda.gov, 2002. of mental health randomized trials evolved over time. J Clin 36 Center for Drug Evaluation and Research. Study 93202. Applica- Epidemiol 2004; 57: 1124–1130. tion number 21-436. Medical review(s). http://www.fda.gov, 2002. 58 Cohen J. Statistical Power Analysis for the Behavioral Sciences. 37 Center for Drug Evaluation and Research. Study 94202. Applica- Academic Press: New York, 1969. tion number 21-436. Medical review(s). http://www.fda.gov, 2002. 59 Leucht S, Kane JM, Etschel E, Kissling W, Hamann J, Engel RR. 38 Honigfeld G. Clozapine: antipsychotic activity in treatment- Linking the PANSS, BPRS, and CGI: clinical implications. resistant schizophrenics. Adv Ther 1984; 1: 77–97. Neuropsychopharmacology 2006; 31: 2318–2325. 39 Beasley CM, Tollefson GD, Tran P, Satterlee W, Sanger T, Hamilton 60 Lieberman JA, Stroup TS, McEvoy JP, Swartz MS, Rosenheck RA, S et al. Olanzapine versus haloperidol and placebo. Acute phase Perkins DO et al. Effectiveness of antipsychotic drugs in patients results of the American double-blind olanzapine trial. Neuro- with chronic schizophrenia. N Engl J Med 2005; 353: 1209–1223. psychopharmacology 1996a; 14: 111–123. 61 Jones PB, Barnes TR, Davies L, Dunn G, Lloyd H, Hayhurst KP 40 Beasley CM, Sanger T, Satterlee W. Olanzapine versus placebo: et al. Randomized controlled trial of the effect on Quality of Life of results of a double-blind fixed dose olanzapine trial. Psycho- second- vs first-generation antipsychotic drugs in schizophrenia— pharmacology 1996b; 124: 159–167. Cost Utility of the Latest Antipsychotic Drugs in Schizophrenia 41 Beasley CM, Sutton VK, Hamilton SH, Walker DJ, Dossenbach M, Study (CUtLASS 1). Arch Gen Psychiatry 2006; 63: 1079–1086. Taylor CC et al. A double-blind, randomized, placebo-controlled 62 Cole JO. Phenothiazine treatment in acute schizophrenia. Arch trial of olanzapine in the prevention of psychotic relapse. J Clin Gen Psychiatry 1964; 10: 246–261. Psychopharmacol 2003; 23: 582–594. 63 Adams CE, Awad G, Rathbone J. Chlorpromazine versus placebo 42 Corrigan MH, Gallen CC, Bonura ML, Merchant KM. Effectiveness for schizophrenia. In: The Cochrane Library. Wiley and Sons: of the selective D-4 antagonist sonepiprazole in schizophrenia: Chichester, UK, 2007. a placebo-controlled trial. Biol Psychiatry 2004; 55: 445–451. 64 Joy C, Adams CE, Laurie S. Haloperidol versus placebo for schizo- 43 Kryzhanovskaya L, Schulz C, McDougle CJ, Frazier JA, Dittmann phrenia. In: The Cochrane Library. Wiley and Sons: Chichester, R, Robertson-Plouch C et al. A double-blind, placebo-controlled UK, 2006. study of olanzapine in adolescents with schizophrenia. Biol 65 Gilbert P, Harris MJ, McAdams LA. Neuroleptic withdrawal Psychiatry 2006; 59: 224S. in schizophrenic patients. A review of the literature. Arch Gen 44 Borison RL, Arvanitis LA, Miller BG. ICI 204,636 an atypical Psychiatry 1995; 52: 173–188. antipsychotic efficacy and safety in a multicenter placebo- 66 Lieberman JA, Phillips M, Gu H, Stroup S, Zhang P, Kong L et al. controlled trial in patients with schizophrenia. J Clin Psycho- Atypical and conventional antipsychotic drugs in treatment-naive pharmacol 1996; 16: 158–169. first-episode schizophrenia: a 52-week randomized trial of 45 Fabre LF, Arvanitis L, Pultz J, Jones VM, Malick JB, Slotnick VB. clozapine vs chlorpromazine. Neuropsychopharmacology 2003; Seroquelt (ICI 204,636), a novel, atypical antipsychotic: early 28: 995–1003. indication of safety and efficacy in patients with chronic and 67 Robinson DG, Woerner MG, Alvir JMJ, Geisler S, Koreen A, subchronic schizophrenia. Clin Ther 1995; 17: 366–378. Sheitman B et al. Predictors of treatment response from a first 46 Small JG, Hirsch SR, Arvanitis LA, Miller BG, Link CGG, Seroquel episode of schizophrenia or schizoaffective disorder. Am J Study Group. Quetiapine in patients with schizophrenia. A high- Psychiatry 1999; 156: 544–549. and low-dose comparison with placebo. Arch Gen Psychiatry 68 Riedel M, Strassnig M, Muller N, Zwack P, Moller HJ. How 1997; 54: 549–557. representative of everyday clinical populations are schizophrenia 47 Borison RL, Pathiraja AP, Diamond BI, Meibach RC. Risperidone— patients enrolled in clinical trials? Eur Arch Psychiatry Clin clinical safety and efficacy in schizophrenia. Psychopharmacol Neurosci 2005; 255: 143–148. Bull 1992; 28: 213–218. 69 Hofer A, Hummer M, Huber R, Kurz M, Walch T, Fleischhacker 48 Chouinard G, Jones B, Remington G. Canadian placebo-controlled WW. Selection bias in clinical trials with antipsychotics. J Clin study of fixed doses of risperidone and haloperidol in the Psychopharmacol 2000; 20: 699–702.

Molecular Psychiatry Atypical antipsychotics versus placebo S Leucht et al 447 70 Heres S, Davis J, Maino K, Jetzinger E, Kissling W, Leucht S. Why 73 Scherk H, Pajonk FG, Leucht S. Second generation antipsychotics olanzapine beats risperidone, risperidone beats quetiapine, and in the treatment of acute mania: a systematic review and meta- quetiapine beats olanzapine: an exploratory analysis of head- analysis of randomized, controlled trials. Arch Gen Psychiatry to-head comparison studies of second-generation antipsychotics. 2007; 64: 442–455. Am J Psychiatry 2006; 163: 185–194. 74 Cavazzoni PA, Berg PH, Kryzhanovskaya LA, Briggs SD, Roddy 71 Leucht S, Pitschel-Walz G, Abraham D, Kissling W. Efficacy and TE, Tohen M et al. Comparison of treatment-emergent extra- extrapyramidal side-effects of the new antipsychotics olanzapine, pyramidal symptoms in patients with bipolar mania or schizo- quetiapine, risperidone, and sertindole compared to conventional phrenia during olanzapine clinical trials. J Clin Psychiatry 2006; antipsychotics and placebo. A meta-analysis of randomized 67: 107–113. controlled trials. Schizophr Res 1999; 35: 51–68. 75 Cheung BM, Lauder IJ, Lau C-P, Kumana CR. Meta-analysis of large 72 Van Putten T, May RP. Akinetic depression in schizophrenia. Arch randomized controlled trials to evaluate the effect of statins on Gen Psychiatry 1978; 35: 1101–1107. cardiovascular outcomes. Br J Clin Pharmacol 2004; 57: 640–651.

Molecular Psychiatry