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