Long-Term Antipsychotic Treatment in Schizophrenia: Systematic Review and Network Meta-Analysis of Randomised Controlled Trials

Home , Amisulpride, Clopenthixol, Trifluoperazine, Zuclopenthixol

Long-term antipsychotic treatment in schizophrenia: systematic review and network meta-analysis of randomised controlled trials

The Harvard community has made this article openly available. Please share how this access benefits you. Your story matters

Citation Zhao, Ying Jiao, Liang Lin, Monica Teng, Ai Leng Khoo, Lay Beng Soh, Toshiaki A. Furukawa, Ross J. Baldessarini, Boon Peng Lim, and Kang Sim. 2016. “Long-term antipsychotic treatment in schizophrenia: systematic review and network meta-analysis of randomised controlled trials.” BJPsych open 2 (1): 59-66. doi:10.1192/bjpo.bp.115.002576. http://dx.doi.org/10.1192/ bjpo.bp.115.002576.

Published Version doi:10.1192/bjpo.bp.115.002576

Citable link http://nrs.harvard.edu/urn-3:HUL.InstRepos:29408187

Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of- use#LAA BJPsych Open (2016) 2, 59–66. doi: 10.1192/bjpo.bp.115.002576

Long-term antipsychotic treatment in schizophrenia: systematic review and network meta-analysis of randomised controlled trials Ying Jiao Zhao, Liang Lin, Monica Teng, Ai Leng Khoo, Lay Beng Soh, Toshiaki A. Furukawa, Ross J. Baldessarini, Boon Peng Lim and Kang Sim

Background decanoate and trifluoperazine produced more extrapyramidal For treatment of patients diagnosed with schizophrenia, adverse effects than olanzapine or quetiapine; and comparative long-term effectiveness of antipsychotic drugs to olanzapine was associated with more weight gain than reduce relapses when minimising adverse effects is of clinical other agents. interest, hence prompting this review. Conclusions Aims Except for apparent superiority of olanzapine and fluphenazine To evaluate the comparative long-term effectiveness of decanoate over chlorpromazine, most agents showed antipsychotic drugs. intermediate efficacy for relapse prevention and differences Method among them were minor. Typical antipsychotics yielded We systematically searched electronic databases for reports adverse neurological effects, and olanzapine was associated with weight gain. The findings may contribute to evidence- of randomised controlled trials (RCTs) of antipsychotic monotherapy aimed at reducing relapse risks in schizophrenia. based treatment selection for patients with chronic psychotic We conducted network meta-analysis of 18 antipsychotics disorders. and placebo. Declaration of interest Results R.J.B. received grants from the Bruce J. Anderson Foundation Studies of 10 177 patients in 56 reports were included; and the McLean Private Donors Psychopharmacology treatment duration averaged 48 weeks (range 4–156). Research Fund. Olanzapine was significantly more effective than chlorpromazine (odds ratio (OR) 0.35, 95% CI 0.14–0.88) Copyright and usage or haloperidol (OR=0.50, 95% CI 0.30–0.82); and © The Royal College of Psychiatrists 2016. This is an open fluphenazine decanoate was more effective than access article distributed under the terms of the Creative chlorpromazine (OR=0.31, 95% CI 0.11–0.88) in relapse Commons Non-Commercial, No Derivatives (CC BY-NC-ND) reduction. Fluphenazine decanoate, haloperidol, haloperidol licence.

Chronic psychotic disorders diagnosed as schizophrenia are metabolic syndrome.7 An additional, major limitation to all long- severe, idiopathic conditions affecting 26 million people world- term maintenance treatments is lack of sustained adherence – wide and resulting in substantial disability in a majority of cases.1 to them.8 10 Long-acting, injectable antipsychotic agents pro‐ Because of their early onset, chronic course and debilitating mise to improve treatment adherence, but evidence of superior effects, schizophrenia ranks among the top 20 causes of years clinical outcomes with such drugs compared with oral agents is lived with disability.2 The course of the illness varies, but most inconsistent.11,12 patients have a chronic course with erratic exacerbations or Given the pressing need for effective long-term treatments for relapses with repeated hospital admissions, decreased quality of schizophrenia and a growing number of available antipsychotic life and a high economic burden. Successive relapses also are drugs, evidence of the relative merits of individual agents is of associated with progressively declining outcomes. Therefore, great interest. Available reviews of evidence of efficacy and relapse prevention is critical for adequate clinical management of tolerability of antipsychotic agents generally indicate minor and this devastating illness.3,4 variable differences between specific drugs, with the notable – Long-term maintenance treatment with antipsychotic medica- exception of clozapine.6,7,11 14 Such comparisons also are severely tion has become the standard for the treatment of patients limited by the paucity of direct, head-to-head comparisons of diagnosed with schizophrenia, with the aim of limiting sympto- specific agents. Recent developments in methods of meta-analysis matic relapses and disability. Although many effective antipsycho- promise to improve this situation, even without direct compar- tic drugs have been developed since the 1950s, all are limited in isons of specific treatments, based on application of network effectiveness and are associated with a range of potentially serious meta-analysis.15 In contrast to traditional pairwise meta-analyses, adverse effects including neurological, metabolic and cardiovas- network methods allow indirect comparisons between treatments cular problems which complicate their long-term use.5,6 Generally, carried out in different trials, under presumably, if not demon- ‘second-generation’ or ‘atypical’ antipsychotics are better toler- strably, highly comparable conditions.16 We now report on results ated than older antipsychotic agents, at least with regard to some of a network meta-analysis to evaluate a total of 18 orally extrapyramidal neurological symptoms, but sometimes present administered and long-acting injectable (LAI) antipsychotic drug high risks of adverse effects associated with weight gain, including preparations.

59 Zhao et al

Method abnormally elevated serum glucose concentration, hyperprolactinaemia, death and suicide attempts. Relative effect size was based Literature search on categorical outcomes, expressed as odds ratio (OR) with its We first performed a PubMed search to identify recent systematic 95% confidence interval (CI). We also used Surface Under the – Cumulative RAnking curve (SUCRA) values to provide a hier- reviews (2009 2015) evaluating the use of antipsychotics for 21 relapse prevention in patients diagnosed with schizophrenia.17 archy of efficacy and tolerability of the treatments considered. Then, individual reports of randomised controlled trials (RCTs) SUCRA is a transformation of the mean rank representing the n identified in the reviews were added to those identified in probability that a treatment was among the best treatments, n literature searches of the PubMed/Medline and Cochrane Library where ranges from 1 to the total number of treatments (here – databases. Searching was based on applying combinations of the 1 18) in the data network. SUCRA of 1 indicates a treatment is following search terms: schizophrenia, antipsychotics, amisulpride, the most effective and 0 the least effective. aripiprazole, chlorpromazine, flupenthixol decanoate, fluphena- As the network meta-analysis combined both direct and zine decanoate, haloperidol and its decanoate, olanzapine, paliper- indirect comparisons to synthesise treatment effects, we assessed idone, long-acting injectable paliperidone, pipothiazine palmitate, agreement between these two sources of evidence with tests of 22 quetiapine, risperidone, long-acting injectable risperidone, sulpir- inconsistency. A loop-specific test evaluates inconsistency in all ide, trifluoperazine, ziprasidone and zuclopenthixol decanoate (see closed loops formed by three or four treatments within each the search term list in the data supplement). network, by comparing direct with indirect estimates of a specific treatment effect. A design6treatment interaction model evaluates Study selection whether the network as a whole demonstrates inconsistency, by evaluating the agreement of evidence generated by trials with Two reviewers (Y.J.Z. and L.L.) screened titles and abstracts different treatment designs. against pre-defined study inclusion criteria and promising articles We investigated potential sources of heterogeneity among were reviewed as full texts. The degree of agreement as evaluated trials with meta-regression modelling and sensitivity analysis by interrater reliability was 0.91. Any disagreements about reports by taking account of study level characteristics. For meta- selected or data extracted were resolved by consensus with a third, regression, we analysed the impact of follow-up duration (<6, senior co-investigator (K.S.). Trials were eligible if they involved 6–11 or ≥12 months), trial setting (in-patient, out-patient or the use of antipsychotic monotherapy for relapse prevention both), trial quality (presence or absence of high risk of bias in any among initially, clinically stable patients diagnosed with schizo- of the Cochrane risk of bias domains defined above), funding phrenia. We excluded trials conducted in patients with predomi- source (with or without industry sponsorship) and publication nant negative symptoms or known treatment resistance and trials year (published before or after year 2000). For sensitivity analysis, for acute illness. We included head-to-head and placebo-con- we evaluated the impact of excluding trials with shorter follow-up trolled RCTs that were single or double-blinded. We did not duration (<6 months), high doses (equivalent of >10 mg/day of restrict the minimum trial length. haloperidol) and published before 1980.23 Publication bias was Data collection and quality assessment assessed using comparison-adjusted funnel plot. In the absence of publication bias, the comparison-adjusted funnel plot is sym- Data regarding patient characteristics, study design, duration of metric around zero line. This study-specific effect sizes do not test treatments, outcomes (relapse rates, drop-out rates, adverse differ from the respective comparison-specific pooled effect effects) and funding sources were extracted independently by two estimates for comparisons between antipsychotic treatments and reviewers (Y.J.Z. and L.L.) with disagreement resolved through placebo. consensus with the same third senior co-investigator (K.S.; Tables DS1 and DS2 in the data supplement). The degree of agreement as evaluated by interrater reliability was 0.90. Quality assessment of Results included studies was performed using the Cochrane Risk of Bias Tool, considering the following six domains of bias: random Description of included studies sequence generation, allocation concealment, masking of partici- Preliminary searching of 32 systematic reviews yielded a total of pants and personnel, masking of outcome assessors, incomplete 186 potentially usable citations of reports of RCTs; another 2054 18 data and selective reporting. reports (total of 2477) were identified by direct searching of electronic databases (Fig. 1). Based on preliminary screening of Data synthesis and analysis abstracts and detailed review of full reports, 56 reports met study We carried out a random effects network meta-analysis using the inclusion criteria and were included for analysis, as indicated in mvmeta routine in STATA 13 statistical software (StataCorp. the study network of direct and indirect comparisons (Fig. 2, College Station, Texas, USA).19,20 Evidence from both direct reference list in the data supplement). Included trials involved 18 (head-to-head trials) and indirect comparisons (between appar- antipsychotic drug products: amisulpride, aripiprazole, chlorpro- ently similar trials using one or more common comparators) were mazine, flupenthixol decanoate (flupenthixol LAI), fluphenazine combined in the analysis. The primary outcome of interest was decanoate (fluphenazine LAI), haloperidol, haloperidol decanoate relapse as defined in each study at its longest follow-up; and when (haloperidol LAI), olanzapine, paliperidone, paliperidone palmi- relapse was not reported, we used clinically assessed or rating tate (paliperidone LAI), pipothiazine palmitate (pipothiazine LAI), scale-based exacerbation of psychotic symptoms considered in the quetiapine, risperidone, risperidone LAI, sulpiride, trifluoperazine, reports to be clinically significant, or readmission, and verified ziprasidone and zuclopenthixol decanoate (zuclopenthixol LAI). A that outcome measures were sufficiently comparable between total of 10 177 adult participants were represented in the 56 included trials compared. Additional outcomes considered included study trials reported between 1960 and 2014; mean age (range) was 39 withdrawal for any reason as a proxy measure of adherence to (15–82) years, and 64% were men. Mean, nominal study duration treatment, study withdrawal because of inefficacy, study with- was 48 weeks (range 4–156). Most trials were conducted in Europe drawal because of adverse events and adverse events includ‐ and American continents; 7% involved Asians. Out-patients were ing extrapyramidal symptoms (EPS), substantial weight gain, represented in 29 (52%) trials; in-patients in 10 (18%), both in 10

60 Comparative long-term efficacy and safety of antipsychotic treatment

Records (n=2477) identiﬁed through database searching (n=2291) and reviews (n=186) Identiﬁcation

Duplicates removed (n=423)

Initial screening of abstracts (n=2054) Screening

Abstracts not meeting inclusion criteria excluded (n=1948)

Full-text articles assessed for eligibility (n=106) Eligibility

Full-text articles not meeting inclusion criteria excluded (n=50)

Studies included (n=56) Included

Fig. 1 Flowchart of selection of reports for inclusion in study-based PRISMA recommendations (www.prisma-statement.org/statement.htm) to yield 56 study reports included in analyses.

(18%), and setting was not reported in another 7 (12%) trials. studies (47/56, 84%) were nominally double-blind and 21 (38%) Financing was based on pharmaceutical industry support in 23 trials tested and confirmed the success of blinding. In most trials, (41%), public sponsors (11%), no support (25%), and unreported relapse represented an end-point leading to discontinuation from sponsorship (23%). Definitions of ‘relapse’ varied: rating scale-based the study; however, drop-out rates were balanced across interven- criteria (23%), hospital admission (15%), combination of rating scale tions in most studies (mean (range): 34% (0–86%)). No evidence and hospital admission (23%), clinical worsening of symptoms of selective reporting was found in 28 (50%) trials included and (29%), need for a change of medication (5%) and undefined the bias on selective report was judged to be unclear in 20 (36%) methods (5%). trials because of the unavailability of study protocol. Most selective reporting (5/8, 63%) was of adverse events, which some investi- Assessment of potential bias gators reported only when at least 5–10% of participants were Most trials did not provide details about randomisation proce- affected. dures and allocation concealment (Fig. DS1). One study (2% of all reports) had high risk of bias in random sequence generation, Consistency between direct and indirect evidence 10 (18%) appear to have had adequate sequence generation and Loop-specific tests revealed no significant inconsistency in 17 4 (7%) had apparently adequate allocation concealment. Most available loops (formed by three or four treatments) within

61 Zhao et al

Fig. 2 Network of all direct and indirect comparisons between antipsychotics. LAI, long-acting injection.

the data network for relapse prevention (based on the 95% risperidone–risperidone LAI. Data extraction was checked and there CI crossing a null value of 1.00). However, the CI was wide were no apparent variables that differed across these comparisons. for four of the loops: placebo–fluphenazine LAI–trifluoperazine, Based on a design6treatment interaction model, no significant placebo–chlorpromazine–trifluoperazine, flupenthixol LAI–haloperi‐ inconsistency between direct and indirect evidence was identified dol LAI–zuclopenthixol LAI and olanzapine–quetiapine– within the evidence network as a whole (P=0.14).

Fig. 3. Forest plot for relapse rate of antipsychotics compared with placebo. LAI, long-acting injection; OR, odds ratio.

62 Comparative long-term efficacy and safety of antipsychotic treatment

PBO 0.25 0.41 0.61 0.21 0.24 0.46 0.16 0.22 0.49 0.18 0.22 0.50 0.31 0.37 0.77 0.74 0.31 0.07 (0.09,0.66) (0.15,1.14) (0.19,1.92) (0.04,0.98) (0.11,0.54) (0.22,0.93) (0.05,0.47) (0.12,0.42) (0.15,1.62) (0.03,0.92) (0.06,0.76) (0.26,0.95) (0.13,0.76) (0.13,1.09) (0.16,3.76) (0.15,3.66) (0.14,0.68) (0.01,0.42) 0.18 AMI 1.64 2.44 0.84 0.98 1.83 0.64 0.89 1.98 0.71 0.87 2.00 1.26 1.50 3.08 2.95 1.23 0.29 (0.07,0.48) (0.42,6.38) (0.55,10.71) (0.14,5.04) (0.28,3.39) (0.81,4.12) (0.16,2.64) (0.38,2.10) (0.42,9.20) (0.11,4.64) (0.18,4.15) (0.76,5.26) (0.46,3.40) (0.29,5.73) (0.57,16.56) (0.48,18.22) (0.46,3.31) (0.04,2.09) 0.24 1.37 ARI 1.48 0.51 0.60 1.11 0.39 0.54 1.20 0.43 0.53 1.22 0.76 0.91 1.87 1.80 0.75 0.18 (0.09,0.66) (0.35,5.37) (0.32,6.89) (0.08,3.18) (0.16,2.17) (0.34,3.68) (0.09,1.69) (0.17, 1.72) (0.25,5.77) (0.06,2.93) (0.11,2.63) (0.41,3.60) (0.21,2.82) (0.33,2.53) (0.29,12.05) (0.27,11.91) (0.21,2.61) (0.02,1.32) 0.31 1.73 1.27 CHL 0.35 0.40 0.75 0.26 0.37 0.81 0.29 0.36 0.82 0.52 0.62 1.26 1.21 0.50 0.12 (0.15,0.64) (0.52,5.78) (0.37,4.36) (0.05,2.37) (0.10,1.63) (0.20,2.80) (0.05,1.29) (0.10,1.33) (0.15,4.26) (0.04,2.18) (0.07,1.96) (0.22,3.04) (0.12,2.15) (0.13,2.95) (0.18,8.75) (0.17,8.67) (0.13,2.02) (0.01,0.97) 0.13 0.74 0.54 0.43 FLX LAI 1.17 2.18 0.77 1.06 2.35 0.84 1.04 2.38 1.49 1.79 3.67 3.52 1.46 0.34 (0.04,0.45) (0.15,3.56) (0.11,2.62) (0.10,1.77) (0.25,5.50) (0.41,11.56) (0.20,2.93) (0.20,5.48) (0.34,16.47) (0.14,5.24) (0.27,4.08) (0.48,11.92) (0.26,8.61) (0.28,11.23) (0.42,32.34) (0.41,30.09) (0.26,8.09) (0.08,1.50) 0.10 0.54 0.40 0.31 0.73 FLZ LAI 1.87 0.66 0.91 2.02 0.72 0.89 2.04 1.28 1.53 3.14 3.01 1.25 0.29 (0.04,0.20) (0.16,1.88) (0.11,1.39) (0.11,0.88) (0.23,2.28) (0.66,5.33) (0.19,2.23) (0.33,2.50) (0.48,8.47) (0.13,4.13) (0.29,2.77) (0.75,5.58) (0.39,4.16) (0.41,5.75) (0.56,17.48) (0.61,14.84) (0.41,3.83) (0.05,1.80) 0.21 1.21 0.88 0.70 1.63 2.23 HAL 0.35 0.49 1.08 0.39 0.48 1.09 0.69 0.82 1.68 1.61 0.67 0.16 (0.11,0.42) (0.51,2.85) (0.28,2.80) (0.27,1.77) (0.41,6.48) (0.82,6.05) (0.10,1.23) (0.29,0.80) (0.27,4.32) (0.07,2.26) (0.12,1.96) (0.52,2.31) (0.37,1.28) (0.25,2.68) (0.38,7.45) (0.30,8.65) (0.30,1.48) (0.02,1.02) 0.15 0.82 0.60 0.47 1.11 1.52 0.68 HAL LAI 1.38 3.07 1.10 1.36 3.11 1.95 2.33 4.79 4.59 1.91 0.45 (0.06,0.35) (0.22,3.03) (0.16,2.24) (0.15,1.48) (0.33,3.75) (0.53,4.31) (0.23,2.00) (0.41,4.69) (0.61,15.39) (0.32,3.81) (0.38,4.89) (0.98,9.88) (0.50,7.61) (0.59,11.00) (0.72,31.69) (0.70,30.28) (0.51,7.07) (0.10,2.00) 0.11 0.60 0.44 0.35 0.81 1.11 0.50 0.73 OLA 2.22 0.80 0.98 2.25 1.41 1.68 3.46 3.32 1.38 0.32 (0.06,0.20) (0.24,1.49) (0.14,1.37) (0.14,0.88) (0.21,3.19) (0.42,2.96) (0.30,0.82) (0.25,2.11) (0.57,8.61) (0.14,4.53) (0.25,3.91) (1.13,4.46) (0.70,2.84) (0.54,5.30) (0.74,16.26) (0.62,17.76) (0.67,2.85) (0.05,2.06) 0.12 0.65 0.48 0.38 0.88 1.21 0.54 0.80 1.09 PAL 0.36 0.44 1.01 0.64 0.76 1.56 2.06 0.62 0.15 (0.04,0.31) (0.16,2.66) (0.12,1.95) (0.11,1.29) (0.18,4.23) (0.35,4.17) (0.17,1.78) (0.21,2.97) (0.34,3.50) (0.05,2.74) (0.08,2.48) (0.26,3.93) (0.14,2.81) (0.15,3.76) (0.21,11.36) (0.23,18.48) (0.15,2.61) (0.02,1.22) 0.19 1.08 0.79 0.62 1.45 1.99 0.89 1.31 1.78 1.64 PAL LAI 1.23 2.82 1.77 2.11 4.34 4.16 1.73 0.41 (0.07,0.50) (0.27,4.28) (0.20,3.15) (0.18,2.08) (0.35,6.07) (0.6106.44) (0.28,2.86) (0.49,3.46) (0.57,5.61) (0.41,6.53) (0.21,7.32) (0.52,15.37) (0.28,11.15) (0.31,14.40) (0.45,41.60) (0.44,39.78) (0.28,11.15) (0.06,2.83) 0.11 0.64 0.47 0.37 0.87 1.19 0.53 0.78 1.07 0.98 0.60 PIP LAI 2.29 1.44 1.72 3.53 3.38 1.40 0.33 (0.04,0.35) (0.14,2.87) (0.10,2.12) (0.10,1.41) (0.31,2.43) (0.45,3.16) (0.15,1.95) (0.24,2.56) (0.30,3.85) (0.22,4.38) (0.15,2.37) (0.59,8.93) (0.32,6.51) (0.34,8.66) (0.49,25.41) (0.50,22.82) (0.32,6.08) (0.06,1.90) 0.21 1.16 0.85 0.67 1.56 2.14 0.96 1.41 1.92 1.77 1.08 1.80 QUE 0.63 0.75 1.54 1.47 0.61 0.14 (0.11,0.39) (0.42,3.23) (0.29,2.49) (0.26,1.75) (0.4076.03) (0.80,5.73) (0.45,2.05) (0.51,3.91) (0.94,3.94) (0.55,5.76) (0.35,3.35) (0.50,6.43) (0.25,1.57) (0.28,2.03) (0.30,7.81) (0.27,8.04) (0.26,1.43) (0.02,0.88) 0.15 0.86 0.63 0.50 1.16 1.59 0.71 1.05 1.43 1.31 0.80 1.33 0.74 RIS 1.20 2.45 2.35 0.98 0.23 (0.07,0.33) (0.32,2.29) (0.19,2.12) (0.18,1.39) (0.27,4.90) (0.54,4.68) (0.42,1.21) (0.33,3.31) (0.78,2.59) (0.37,4.61) (0.23,2.75) (0.34,5.19) (0.32,1.74) (0.33,4.38) (0.49,12.26) (0.40,13.81) (0.38,2.54) (0.03,1.61) 0.17 0.94 0.69 0.54 1.27 1.74 0.78 1.14 1.56 1.44 0.87 1.46 0.81 1.09 RIS LAI 2.05 1.97 0.82 0.19 (0.06,0.47) (0.25,3.58) (0.25,1.87) (0.15,1.90) (0.26,6.21) (0.48,6.22) (0.25,2.41) (0.31,4.29) (0.52,4.72) (0.34,6.01) (0.22,3.55) (0.32,6.64) (0.31,2.15) (0.34,3.57) (0.32,13.20) (0.29,13.27) (0.23,2.85) (0.03,1.45) 0.16 0.90 0.66 0.52 1.22 1.67 0.75 1.10 1.50 1.38 0.84 1.40 0.78 1.05 0.96 SUL 0.96 0.40 0.09 (0.03,0.91) (0.14,5.82) (0.09,4.80) (0.08,3.24) (0.15,10.06) (0.25,10.89) (0.14,3.96) (0.16,7.60) (0.27,8.37) (0.19,10.18) (0.12,6.11) (0.18,10.92) (0.13,4.68) (0.18,5.99) (0.13,6.92) (0.15,6.27) (0.08,2.09) (0.01,0.97) 0.28 1.60 1.17 0.93 2.16 2.96 1.33 1.95 2.66 2.45 1.49 2.49 1.38 1.87 1.71 1.78 TRI 0.42 0.10 (0.06,1.25) (0.28,9.18) (0.20,6.98) (0.21,4.09) (0.33,14.17) (0.60,14.55) (0.28,6.36) (0.36,10.73) (0.55,12.82) (0.41,14.51) (0.26,8.69) (0.41,15.20) (0.28,6.84) (0.37,9.49) (0.28,10.25) (0.25,12.74) (0.07,2.040) (0.01,1.00) 0.13 0.74 0.54 0.43 1.00 1.37 0.62 0.91 1.24 1.14 0.69 1.16 0.64 0.87 0.79 0.83 0.46 ZIP 0.23 (0.06,0.31) (0.25,2.25) (0.15,1.95) (0.14,1.28) (0.23,4.39) (0.45,4.24) (0.26,1.46) (0.27,2.99) (0.55,2.76) (0.31,4.13) (0.19,2.48) (0.29,4.66) (0.25,1.64) (0.34,2.24) (0.22,2.81) (0.13,5.24) (0.09,2.48) (0.03,1.59) 0.05 0.31 0.22 0.18 0.41 0.57 0.25 0.37 0.51 0.47 0.29 0.48 0.27 0.36 0.33 0.34 0.19 0.41 ZUC LAI (0.01,0.38) (0.03,2.71) (0.03,2.00) (0.02,1.42) (0.08,2.27) (0.08,3.91) (0.03,1.98) (0.06,2.47) (0.07,3.92) (0.05,4.15) (0.04,2.24) (0.07,3.21) (0.03,2.01) (0.04,2.89) (0.04,2.93) (0.03,4.59) (0.02,2.15) (0.05,3.40)

Fig. 4. Relapse rate and discontinuation because of all reasons for all antipsychotics.

AMI, amisulpride; ARI, aripiprazole; CHL, chlorpromazine; FLX, flupenthixol decanoate; FLZ, fluphenazine decanoate; HAL, haloperidol; LAI, long-acting injection; OLA, olanzapine; PAL, paliperidone; PBO, placebo; PIP, pipothiazine palmitate; QUE, quetiapine; RIS, risperidone; SUL, sulpiride; TRI, trifluoperazine; ZIP, ziprasidone; ZUC, zuclopenthixol decanoate.

Efficacy ranked by their SUCRA values were zuclopenthixol LAI (0.85), All antipsychotic treatments were significantly better than placebo fluphenazine LAI (0.78), olanzapine (0.76), pipothiazine LAI (0.68) for relapse prevention, with the single and surprising exception of and paliperidone (0.67). Findings for relapse prevention were trifluoperazine. However, most CIs overlapped between treatments consistent when we analysed oral and long-acting agents separately so that few comparisons were different in statistical significance (Fig. DS2). (Figs. 3 and 4). Olanzapine was significantly more effective than Fourteen trials (2886 participants) were included for the chlorpromazine (OR=0.35, 95% CI 0.14–0.88) and haloperidol network meta-analysis of hospital admissions. Oral agents includ- (OR=0.50, 95% CI 0.30–0.82) in reducing relapses, and fluphenazine ing amisulpride, haloperidol, olanzapine, quetiapine and ziprasi- LAI was superior to chlorpromazine (OR=0.31, 95% CI done were significantly more effective in reducing readmissions 0.11–0.88; Fig. 4). Differences between other pairs of drugs were than placebo, but aripiprazole, chlorpromazine, paliperidone, minor and not significant. We also evaluated hierarchies of efficacy trifluoperazine did not differ from placebo (Fig. DS3). There was on the basis of SUCRA rankings for relapse rate. The top five drugs a lack of data on hospital admission rates with long-acting agents

PBO 0.99 AMI (0.50,1.97) 0.87 0.88 ARI (0.38,1.98) (0.34,2.24) 1.28 1.29 1.47 CHL (0.67,2.43) (0.53,3.16) (0.51,4.19) 1.53 1.54 1.76 1.20 FLX LAI (0.20,11.54) (0.18,13.04) (0.20,15.52) (0.14,10.00) 4.63 4.68 5.32 3.62 3.03 FLZ LAI (1.22,17.51) (1.05,20.87) (1.12,25.34) (0.82,15.91) (0.32,28.55) 2.34 2.36 2.69 1.83 1.53 0.51 HAL (1.28,4.29) (1.50,3.71) (1.10,6.57) (0.80,4.21) (0.19,12.60) (0.12,2.18) 6.02 6.08 6.91 4.71 3.93 1.30 2.57 HAL LAI (1.71,21.19) (1.45,25.47) (1.54,30.96) (1.14,19.44) (0.81,19.12) (0.26,6.39) (0.64,10.39) 0.51 0.51 0.58 0.40 0.33 0.11 0.22 0.08 OLA (0.26,0.97) (0.29,0.89) (0.23,1.45) (0.16,0.97) (0.04,2.76) (0.02,0.48) (0.14,0.33) (0.02,0.35) 2.40 2.43 2.76 1.88 1.57 0.52 1.03 0.40 4.75 PAL (0.86,6.67) (0.71,8.30) (0.74,10.23) (0.56,6.28) (0.16,15.11) (0.10,2.78) (0.31,3.37) (0.08,2.02) (1.41,16.00) 2.89 2.92 3.32 2.26 1.89 0.63 1.24 0.48 5.73 1.21 PAL LAI (0.95,8.82) (0.79,10.81) (0.83,13.25) (0.62,8.21) (0.34,10.63) (0.13,2.95) (0.35,4.40) (0.24,0.96) (1.58,20.84) (0.27,5.47) 4.66 4.70 5.35 3.64 3.05 1.01 1.99 0.77 9.22 1.94 1.61 PIP LAI (0.97,22.26) (0.85,25.94) (1.12,25.34) (0.67,19.88) (0.41,22.76) (0.20,5.06) (0.37,10.65) (0.22,2.68) (1.70,50.07) (0.30,12.57) (0.41,6.27) 0.69 0.70 0.80 0.54 0.45 0.15 0.30 0.12 1.37 0.29 0.24 0.15 QUE (0.39,1.24) (0.37,1.32) (0.37,1.70) (0.23,1.26) (0.06,3.72) (0.04,0.64) (0.16,0.53) (0.03,0.46) (0.73,2.59) (0.09,0.94) (0.07,0.84) (0.03,0.79) 0.96 0.97 1.10 0.75 0.63 0.21 0.41 0.16 1.90 0.40 0.33 0.21 1.38 RIS (0.49,1.88) (0.56,1.67) (0.43,2.79) (0.31,1.84) (0.07,5.25) (0.05,0.92) (0.29,0.58) (0.04,0.66) (1.27,2.83) (0.12,1.35) (0.09,1.21) (0.04,1.12) (0.72,2.65) 1.19 1.21 1.37 0.93 0.78 0.26 0.51 0.20 2.36 0.50 0.41 0.26 1.72 1.25 RIS LAI (0.55,2.58) (0.51,2.85) (0.80,2.36) (0.35,2.52) (0.09,6.78) (0.06,1.20) (0.23,1.16) (0.05,0.87) (1.01,5.52) (0.14,1.79) (0.11,1.60) (0.04,1.46) (0.92,3.23) (0.53,2.05) 7.38 7.45 8.47 5.77 4.82 1.59 3.15 1.23 14.61 3.07 2.55 1.58 10.64 7.71 6.18 TRI (1.51,21.19) (1.34,41.46) (1.41,50.86) (1.23,27.17) (0.37,63.21) (0.20,12.68) (0.59,17.00) (0.16,9.34) (2.64,80.88) (0.46,20.33) (0.37,17.78) (0.17,14.78) (1.97,57.29) (1.39,42.81) (1.06,36.01) 1.45 1.46 1.66 1.13 0.95 0.31 0.62 0.24 2.86 0.60 0.50 0.31 2.09 1.51 1.21 0.20 ZIP (0.20,2.92) (0.77,2.79) (0.64,4.33) (0.45,2.86) (0.11,8.02) (0.07,1.41) (0.34,1.14) (0.06,1.02) (1.51,5.44) (0.17,2.08) (0.13,1.87) (0.06,1.72) (1.06,4.12) (0.78,2.94) (0.50,2.95) (0.03,1.11) 2.68 2.70 3.07 2.09 1.75 0.58 1.14 0.44 5.30 1.11 0.92 0.57 3.86 2.80 2.24 0.36 1.85 ZUC LAI (0.49,14.49) (0.44,16.72) (0.47,20.05) (0.34,12.80) (0.58,5.31) (0.08,4.06) (0.19,6.88) (0.14,1.37) (0.87,32.33) (0.15,8.02) (0.25,3.47) (0.11,3.08) (0.65,23.03) (0.45,17.18) (0.35,14.33) (0.04,3.70) (0.30,11.51)

Fig. 5. Extrapyramidal symptoms in the context of antipsychotic use.

63 Zhao et al

except for fluphenazine LAI and risperidone LAI, which both demonstrated a consistent trend in reducing relapse rate compared reduced readmission risk more than placebo. In reducing read- with the primary analysis (Fig. DS8a). By excluding trials with mission, olanzapine and ziprasidone were significantly more relatively high doses of haloperidol (>10 mg/day), the treatment effective than haloperidol or quetiapine; amisulpride, fluphenazine effects in relapse prevention remained largely unchanged (Fig. LAI, haloperidol, olanzapine, risperidone LAI and ziprasidone all DS8b). By excluding trials published before 1980, sulpiride and were superior to aripiprazole, and ziprasidone alone outperformed trifluoperazine were excluded from the analysis and other treat- chlorpromazine and trifluoperazine. ments demonstrated a consistent trend in preventing relapse compared with the primary analysis (Fig. DS8c). There was no Safety and tolerability evidence of publication bias given that comparison-adjusted funnel Most drugs were less associated with all-cause discontinuation plot was symmetric around the zero line (Fig. DS9). than placebo, except for aripiprazole, chlorpromazine, paliperidone, risperidone LAI, sulpiride and trifluoperazine. In general, long-acting agents tended to be better tolerated than oral agents, Discussion but not statistically significant (Fig. 4). Olanzapine was associated with less all-cause discontinuation than quetiapine (OR=0.44, 95% This study compared effectiveness of 18 antipsychotics in CI 0.22–0.88) or haloperidol (OR=0.49, 95% CI 0.29–0.80), preventing relapses among initially stable patients with schizo- whereas zuclopenthixol LAI yielded less all-cause discontinua‐ phrenia and provided evidence-based hierarchies using network tion than chlorpromazine (OR=0.12 (0.01–0.97)), quetiapine meta-analysis. This method can extend existing data derived from (OR=0.14 (0.02–0.88)) or sulpiride (OR=0.09 (0.01–0.97)). In traditional pairwise meta-analyses and help to overcome limited separate analyses of discontinuation because of lack of efficacy or availability of head-to-head comparisons of most psychotropic 15 adverse events, most antipsychotics demonstrated lower disconti- treatments. The present findings support the conclusion that nuation rates because of lack of efficacy compared with placebo, maintenance treatment with most clinically available antipsycho- whereas no drug demonstrated a significant difference in disconti- tics reduced relapse rate in patients with schizophrenia over time, nuation rates because of adverse events in comparison with placebo averaging 1 year. Olanzapine was significantly more effective than (Fig. DS4). chlorpromazine or haloperidol, and fluphenazine decanoate Thirty trials (7381 participants) were included for the analysis yielded significant lower relapse rate than chlorpromazine. Of of EPS. Only fluphenazine LAI, haloperidol, haloperidol LAI the top five drugs, ranked by their SUCRA values for relapse and trifluoperazine, but not other antipsychotics, were associated prevention, three were LAI preparations: fluphenazine decanoate, with more reported EPS than placebo (Fig. 5). Olanzapine olanzapine, paliperidone, pipothiazine palmitate and zuclo- was associated with less risk of EPS than other agents except penthixol decanoate. These findings corroborate the results of aripiprazole, flupenthixol LAI, quetiapine and zuclopenthixol LAI. earlier reviews which found olanzapine to be significantly superior As expected, quetiapine had less reported EPS than fluphenazine to older, typical antipsychotic agents in relapse prevention within 7 LAI, haloperidol, haloperidol LAI, paliperidone, paliperidone LAI, schizophrenia, and found fluphenazine decanoate to be more 11 pipothiazine LAI, trifluoperazine and ziprasidone. Fluphenazine effective than several oral antipsychotics. Although zuclo- LAI, haloperidol, haloperidol LAI and trifluoperazine were penthixol decanoate was associated with apparently particularly associated with significantly more EPS than several other agents. favourable effects on relapse prevention, it did not differ Only 15 trials (5147 participants) were synthesised for weight statistically from other agents, based on findings of only two trials 24,25 gain. Olanzapine produced significantly more weight gain than involving fewer than 40 participants per trial. Additionally, the amisulpride, haloperidol, quetiapine, risperidone, ziprasidone and apparent superiority of sulpiride and inferiority of trifluoperazine placebo. Ziprasidone was associated with less weight gain than compared with placebo need to be interpreted with caution amisulpride, quetiapine and risperidone (Fig. DS5). because of relatively wide CIs associated with their treatment Amisulpride, haloperidol, olanzapine, quetiapine, ziprasidone effects. and paliperidone LAI were not associated with higher rate of With respect to readmission rates, olanzapine was also found glucose intolerance than placebo or as compared with each other. to be significantly more effective than aripiprazole, haloperidol or Aripiprazole, paliperidone and paliperidone LAI did not lead to quetiapine. Also, ziprasidone was associated with lower read- more hyperprolactinaemia than placebo, whereas amisulpride, mission rates than aripiprazole, chlorpromazine, haloperidol, risperidone and risperidone LAI produced hyperprolactinaemia quetiapine or trifluoperazine, which were consistent with other more often than haloperidol, olanzapine, quetiapine or ziprasi- evidence of superiority of ziprasidone to some first-generation 7 done (Fig. DS6). Additionally, we found no differences between antipsychotics, at least, in preventing readmission. antipsychotics and placebo or among antipsychotics in terms of LAI agents tended generally to be less likely to be discontinued death or suicide attempt (Fig. DS7). prematurely for all causes (intolerability or inefficacy) than oral antipsychotics, but most of these comparisons were not Meta-regression analysis and sensitivity analysis statistically significant. However, zuclopenthixol decanoate was Treatment effects in reducing relapse rate did not differ signifi- significantly less associated with all-cause discontinuation than cantly (P>0.05) with length of follow-up, trial setting, quality of the quetiapine or trifluoperazine, but it is not clear whether these trials, funding source and publication year as assessed in network differences reflect superior benefit or better tolerability. Similarly, meta-regressions. When analysing the association between relapse olanzapine had significantly lower all-cause discontinuation rates rate and trial duration regardless of the treatment administered, than other agents including haloperidol and quetiapine. As we found that relapse rate was strongly related to nominal expected, most drugs were associated with less discontinuation trial duration (t=6.00, P<0.0001). Accordingly, we carried out a because of lack of efficacy than placebo but such differences were sensitivity analysis that excluded trials with nominal trial follow-up not sustained for adverse events. for <6 months, the difference in relapse prevention between In general, as expected, older, typical antipsychotics were fluphenazine LAI and chlorpromazine became insignificant (OR= associated with more EPS than modern, atypical agents or 0.39, 95% CI 0.12–1.21). Other treatments except sulpiride placebo.7 However, this association was driven by only some excluded from the analysis as its two trials involved brief exposures antipsychotics: fluphenazine decanoate, haloperidol, haloperidol

64 Comparative long-term efficacy and safety of antipsychotic treatment

decanoate and trifluoperazine, whereas others, including chlor- needs to be taken when we interpret the relative effectiveness promazine, flupenthixol decanoate, pipothiazine palmitate and between oral and long-acting antipsychotics. Most adverse effects zuclopenthixol decanoate, were not associated with more EPS than reported appear to have arisen from typically passive and placebo. Among atypical antipsychotics, olanzapine and quetia- incidental reports rather than deliberate, systematic and protocol- pine had less risk of EPS than most other modern agents guided assessments. Moreover, the findings on adverse events such evaluated. However, atypical agents, on average, and especially as glucose intolerance, hyperprolactinaemia and death or suicide olanzapine, produced more weight gain than placebo. Olanzapine attempt need to be interpreted with caution given the limited also produced significantly more weight gain than amisulpride, amount of evidence available. Ideally, a larger and more detailed haloperidol, quetiapine, risperidone and ziprasidone, whereas data-set would allow better examination of their relative effects. It ziprasidone was associated with less weight gain than amisulpride, must be pointed out, however, that most of these limitations are quetiapine and risperidone. These findings accord with previous found in similar reviews and meta-analyses, be they pairwise or research indicating that olanzapine and quetiapine have low risks network meta-analyses. of EPS, but are more likely to lead to weight gain and metabolic In conclusion, relatively minor differences in relapse preven- syndrome,26 whereas ziprasidone had relatively low risks of both tion were observed among most antipsychotics, although olanza- EPS and weight gain.27 pine and fluphenazine decanoate were associated with particularly Amisulpride, aripiprazole, haloperidol, paliperidone, paliper- lower relapse rates. These relative apparent benefits need to be idone palmitate, quetiapine and ziprasidone were associated with weighed against the risks of adverse effects of all antipsychotic similarly low rates of hyperglycaemia as with placebo. Risk of drugs, notably of weight gain and metabolic syndrome with hyperprolactinaemia was highest with amisulpride, risperidone olanzapine, and EPS with fluphenazine decanoate. and risperidone LAI, consistent with previous reviews.28,29 Also, there were no significant differences among drugs or between Ying Jiao Zhao, PhD, Liang Lin, MSc (IHTA), Monica Teng, MHSc, Ai Leng Khoo, antipsychotics and placebo in rates of suicidal behaviours or death PhD, Pharmacy and Therapeutics Office, Group Corporate Development, National from all causes, although studies with such information were few Healthcare Group, Singapore, Singapore; Lay Beng Soh, BPharm, Department of Pharmacy, Institute of Mental Health, Singapore, Singapore; Toshiaki A. Furukawa, and their duration averaged less than a year. Finally, treatment MD, Department of Health Promotion and Human Behavior, Graduate School of discontinuation, in turn, represents a major contribution to risk of Medicine, Kyoto University, Kyoto, Japan; Department of Clinical Epidemiology, relapse, readmission and poor clinical outcomes in psychotic Graduate School of Public Health, Kyoto University, Kyoto, Japan; Ross J. Baldessarini, 8–10 MD, Department of Psychiatry, Harvard Medical School, Boston, Massachusetts, USA, disorders. International Consortium for Psychotic and Mood Disorders Research, McLean This is the first study using network meta-analysis to compare Hospital, Belmont, Massachusetts, USA; Boon Peng Lim, BPharm, Pharmacy and Therapeutics Office, Group Corporate Development, National Healthcare Group, the comparative efficacy and tolerability of antipsychotic mono- Singapore, Singapore; Kang Sim, MBBS, MMed (Psychiatry), FAMS, Department of therapy aimed at reducing rates of relapse or readmission in General Psychiatry, Institute of Mental Health, Singapore, Singapore; Yong Loo Lin patients diagnosed with schizophrenia. The method of network School of Medicine, National University of Singapore, Singapore, Singapore meta-analysis can overcome the lack of head-to-head clinical trials Correspondence: Ying Jiao Zhao, 3 Fusionopolis Link, #03-08 Nexus@one-north, in the search for evidence-based hierarchies of comparative Singapore 138543. Email: [email protected] treatment effects, provided that the underlying assumptions First received 9 Dec 2015, accepted 13 Dec 2015 (notably very close similarity of compared trials) of the method are met.16 We considered 18 antipsychotic drug preparations that are commonly used internationally for maintenance treatment of patients with schizophrenia, and considered several outcome measures related to relapse, including risk of psychiatric hospital References admission, which is one of the main contributors to the cost of care in schizophrenia. We also compared drugs for their risks of 1 Eaton WW, Martins SS, Nestadt G, Bienvenu OJ, Clarke D, Alexandre P. The burden early discontinuation because of various causes, as well as for of mental disorders. Epidemiol Rev 2008; 30:1–14. adverse events severe enough to produce treatment discontinua- 2 Vos T, Flaxman AD, Naghavi M, Lozano R, Michaud C, Ezzati M, et al. Years lived tion and which tend to counterbalance the benefits of treatment. with disability (YLDs) for 1160 sequelae of 289 diseases and injuries 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 2012; The results of this study should be interpreted in the context of 380: 2163–96. its limitations. It is difficult with any data network investigating – 3 Andreasen NC. Symptoms, signs, and diagnosis of schizophrenia. Lancet 1995; indirect drug drug comparisons by network meta-analytic meth- 346: 477–81. ods to consistently compare perfectly matched trials. In the present 4 Wyatt RJ. Neuroleptics and the natural course of schizophrenia. Schizophr Bull analyses, notably the range of treatment exposure time ranged 1991; 17: 325–51. ‘ ’ from 1 month to 2 years, criteria for relapse varied greatly among 5 Rettenbacher MA, Hofer A, Eder U, Hummer M, Kemmler G, Weiss EM, et al. trials and were not always matched in the comparisons reported. In Compliance in schizophrenia: psychopathology, side effects, and patients’ addition, 44% (8/18) of the agents were represented in only 2–3 attitudes toward the illness and medication. J Clin Psychiatry 2004; 65: 1211–18. trials and placebo controls were available for only 45% (25/56) of 6 Leucht S, Tardy M, Komossa K, Heres S, Kissling W, Davis JM. Maintenance trials. Less certain is how broadly the present findings might treatment with antipsychotic drugs for schizophrenia. Cochrane Database Syst Rev 2012; 5: CD008016. pertain to different types of typically heterogeneous patients with 7 Kishimoto T, Agarwal V, Kishi T, Leucht S, Kane JM, Correll CU. Relapse prevention chronic psychotic disorders that meet standard diagnostic criteria in schizophrenia: a systematic review and meta-analysis of second-generation for schizophrenia. That is, it was not possible with the available antipsychotics versus first-generation antipsychotics. Mol Psychiatry 2013; 18: data to consider potential effects of such clinical factors as age at 53–66. onset or current age, years of illness, rates of previous hospital 8 Lacro JP, Dunn LB, Dolder CR, Leckband SG, Jeste DV. Prevalence of and risk admission, diagnostic subtypes, prominence of particular symp- factors for medication nonadherence in patients with schizophrenia: a compre- hensive review of recent literature. J Clin Psychiatry 2002; 63: 892–909. toms, cognitive and social impairment or other functional 9 Velligan DI, Wang M, Diamond P, Glahn DC, Castillo D, Bendle S, et al. Relation- measures, or even to compare changes in standardised symptom ships among subjective and objective measures of adherence to oral antipsycho- rating scales. Individuals also tend to be more adherent in clinical tic medications. Psychiatr Serv 2007; 58: 1187–92. trials than real-world setting that may diminish the difference 10 Kane JM, Garcia-Ribera C. Clinical guideline recommendations for antipsychotic between oral and long-acting antipsychotics. Therefore, caution long-acting injections. Br J Psychiatry Suppl 2009; 52: S63–7.

65 Zhao et al

11 Kishimoto T, Robenzadeh A, Leucht C, Leucht S, Watanabe K, Mimura M, et al. 22 Singh S, Cuzick J, Mesher D, Richmond B, Howell A. Effect of baseline serum Long-acting injectable vs oral antipsychotics for relapse prevention in schizo- vitamin D levels on aromatase inhibitors induced musculoskeletal symptoms: phrenia: a meta-analysis of randomized trials. Schizophr Bull 2014; 40: 192–213. results from the IBIS-II, chemoprevention study using anastrozole. Breast Cancer Res Treat 2012; 132: 625–9. 12 Hartling L, Abou-Setta AM, Dursun S, Mousavi SS, Pasichnyk D, Newton AS. Antipsychotics in adults with schizophrenia: comparative effectiveness of first- 23 Gardner DM, Murphy AL, O’Donnell H, Centorrino F, Baldessarini RJ. International generation versus second-generation medications: a systematic review and meta- consensus study of antipsychotic dosing. Am J Psychiatry 2010; 167: 686–93. analysis. Ann Intern Med 2012; 157: 498–511. 24 Dencker SJ, Lepp M, Malm U. Clopenthixol and flupenthixol depot preparations in 13 Essali A, Al-Haj Haasan N, Li C, Rathbone J. Clozapine versus typical neuroleptic outpatient schizophrenics. I. A one year double-blind study of clopenthixol Cochrane Database Syst Rev medication for schizophrenia. 2009; 1: CD000059. decanoate and flupenthixol palmitate. Acta Psychiatr Scand Suppl 1980; 279: 14 Asenjo Lobos C, Komossa K, Rummel-Kluge C, Hunger H, Schmid F, Schwarz S, 10–28. et al. Clozapine versus other atypical antipsychotics for schizophrenia. Cochrane 25 Wistedt B, Koskinen T, Thelander S, Nerdrum T, Pedersen V, Molbjerg C. Database Syst Rev 2010; 11: CD006633. Zuclopenthixol decanoate and haloperidol decanoate in chronic schizophrenia: 15 Caldwell DM, Ades AE, Higgins JP. Simultaneous comparison of multiple a double-blind multicentre study. Acta Psychiatr Scand 1991; 84:14–21. treatments: combining direct and indirect evidence. BMJ 2005; 331: 897–900. 26 Komossa K, Rummel-Kluge C, Hunger H, Schmid F, Schwarz S, Duggan L, et al. 16 Yildiz A, Vieta E, Correll CU, Nikodem M, Baldessarini RJ. Critical issues on the use Olanzapine versus other atypical antipsychotics for schizophrenia. Cochrane of network meta-analysis in psychiatry. Harv Rev Psychiatry 2014; 22: 367–72. Database Syst Reviews 2010; 3: CD006654. 17 Ziakas PD, Kourbeti IS, Mylonakis E. Systemic antifungal prophylaxis after 27 Komossa K, Rummel-Kluge C, Hunger H, Schwarz S, Bhoopathi PS, Kissling W, hematopoietic stem cell transplantation: a meta-analysis. Clin Ther 2014; 36: et al. Ziprasidone versus other atypical antipsychotics for schizophrenia. 292–306 e1. Cochrane Database Syst Rev 2009; 4: CD006627. 18 Simmonds MC, Higgins JP, Stewart LA, Tierney JF, Clarke MJ, Thompson SG. Meta- 28 Rummel-Kluge C, Komossa K, Schwarz S, Hunger H, Schmid F, Lobos CA, et al. analysis of individual patient data from randomized trials: a review of methods Head-to-head comparisons of metabolic side effects of second generation Clin Trials – used in practice. 2005; 2: 209 17. antipsychotics in the treatment of schizophrenia: a systematic review and meta- Schizophr Res – 19 Mavridis D, Salanti G. A practical introduction to multivariate meta-analysis. Stat analysis. 2010; 123: 225 33. Methods Med Res – 2013; 22: 133 58. 29 Komossa K, Rummel-Kluge C, Schwarz S, Schmid F, Hunger H, Kissling W, et al. 20 Jackson D, White IR, Riley RD. A matrix-based method of moments for fitting the Risperidone versus other atypical antipsychotics for schizophrenia. Cochrane multivariate random effects model for meta-analysis and meta-regression. Biom J Database Syst Rev 2011; 1: CD006626. 2013; 55: 231–45. 21 Chaimani A, Higgins JP, Mavridis D, Spyridonos P, Salanti G. Graphical tools for network meta-analysis in STATA. PLoS One 2013; 8: e76654.