Few controlled trials compared second-generation antipsychotics (SGAs) with first-generation antipsychotics (FGAs) regarding relapse prevention in schizophrenia. We conducted a systematic review/meta-analysis of randomized trials, lasting ⩾6 months comparing SGAs with FGAs in schizophrenia. Primary outcome was study-defined relapse; secondary outcomes included relapse at 3, 6 and 12 months; treatment failure; hospitalization; and dropout owing to any cause, non-adherence and intolerability. Pooled relative risk (RR) (±95% confidence intervals (CIs)) was calculated using random-effects model, with numbers-needed-to-treat (NNT) calculations where appropriate. Across 23 studies (n=4504, mean duration=61.9±22.4 weeks), none of the individual SGAs outperformed FGAs (mainly haloperidol) regarding study-defined relapse, except for isolated, single trial-based superiority, and except for risperidone's superiority at 3 and 6 months when requiring ≥3 trials. Grouped together, however, SGAs prevented relapse more than FGAs (29.0 versus 37.5%, RR=0.80, CI: 0.70–0.91, P=0.0007, I2=37%; NNT=17, CI: 10–50, P=0.003). SGAs were also superior regarding relapse at 3, 6 and 12 months (P=0.04, P<0.0001, P=0.0001), treatment failure (P=0.003) and hospitalization (P=0.004). SGAs showed trend-level superiority for dropout owing to intolerability (P=0.05). Superiority of SGAs regarding relapse was modest (NNT=17), but confirmed in double-blind trials, first- and multi-episode patients, using preferentially or exclusively raw or estimated relapse rates, and for different haloperidol equivalent comparator doses. There was no significant heterogeneity or publication bias. The relevance of the somewhat greater efficacy of SGAs over FGAs on several key outcomes depends on whether SGAs form a meaningful group and whether mid- or low-potency FGAs differ from haloperidol. Regardless, treatment selection needs to be individualized considering patient- and medication-related factors.
As psychopathology and social functioning can worsen with repeated relapses in schizophrenia patients,1 relapse prevention is a critical issue in managing this illness. Since clozapine, the first second-generation antipsychotic (SGA) introduced in 1971 (marketed in the United States of America in 1990), and risperidone, introduced in 1994, a total of eight additional SGAs are now available in the United States of America, which are widely used.2 SGAs are better tolerated than first-generation antipsychotics (FGAs) regarding acute extra-pyramidal side effects (EPS)3 and tardive dyskinesia.4 However, there is growing concern about metabolic side effects, such as body weight gain, insulin resistance and dyslipidemia.5, 6 Combined with the lack of significant superiority in efficacy and/or effectiveness observed in large, pragmatic trials,7, 8, 9, 10 the advantages of non-clozapine SGAs over FGAs have been challenged. Less attention has been focused on relapse prevention. A meta-analysis comparing SGAs to FGAs was published in 2003,11 but since then, there have been 12 additional relevant trials.
Materials and methods
We conducted a search using MEDLINE/PubMed, the Cochrane library and PsycINFO (last search date January 2011) for randomized, controlled trials of relapse prevention or maintenance treatment of schizophrenia and related disorders lasting ⩾6 months. Studies had to be published in English in peer-reviewed journals. Search terms included antipsychotic(s); neuroleptic(s); individual names of SGAs and FGAs; schizophrenia; random, randomly, randomized; and maintenance, relapse or long-term. The electronic search was supplemented by hand search of reference lists of relevant studies and reviews. Authors and companies were contacted to provide missing information and unpublished data.
The trials included in this analysis were randomized, head-to-head comparisons of oral SGAs versus oral FGAs for relapse prevention or maintenance treatment in adults with schizophrenia. We only included trials with a minimum duration of 6 months (one study included patients with a range of 22–84 weeks’ completion12). We also only included trials providing relapse-related information, such as study-defined relapse or re-hospitalization. Trials were included irrespective of whether randomization occurred during the acute or maintenance phase. However, when patients were randomized in the acute phase, we only used data from patients for whom information was available after they had responded, remitted or were discharged.
Data extraction and outcomes
Data were extracted independently by ⩾2 reviewers (T Kishimoto, V Agarwal, T Kishi, CU Correll). Any disagreements were resolved by consensus.
The primary outcome measure was study-defined relapse at endpoint, preferentially based on survival curves, which we believe yield more accurate data for relapse than raw relapse rates, as the bias of unequal follow-up duration is minimized. If the estimated relapse rate was not available, we used raw relapse rate. Although we used study-defined relapse, when there was no definition of relapse or the authors’ definition was regarded as inappropriate, we used the next most appropriate outcome for our analysis, which predominantly was re-hospitalization.
As secondary outcomes, relapse rates at 3, 6 and 12 months; ‘treatment failure’ (defined as relapse and/or all-cause discontinuation, depending on whether data were available for both outcomes); hospitalization; and dropout owing to any cause, non-adherence and intolerability were examined.
All outcomes were dichotomous and SGAs were compared to FGAs both individually and as separate groups for each outcome. As a general rule, we applied a “once-randomized-analyzed” endpoint analysis. Modifications to this rule included studies in which only patients reaching the maintenance phase were considered at risk for relapse and studies where the population entered into the Kaplan-Meier survival curves did not include all originally randomized patients. Pooled relative risk (RR) (±95% confidence intervals (CIs)) and risk differences were calculated using random-effects models by DerSimonian and Laird,13 which is more conservative than fixed-effects models. Number-needed-to-treat (NNT) was calculated where appropriate. To reduce a potential type-I error, we considered the meta-analytic results to be significant only if they were based on ⩾3 studies.
Heterogeneity between studies was explored with a χ2-test of homogeneity (P<0.1) together with the I2-statistic, with I2≥50% indicating significant heterogeneity.14
In addition to the primary and secondary outcome analyses, we also conducted a priori defined sensitivity analyses of the primary outcome comparing SGAs with FGAs, seeking to identify potential methodological biases and whether the findings extended to clinically relevant sub-populations and treatment groups. The examined variables included (a) treatment concealment (open versus blinded); (b) sponsorship (industry versus academia); (c) publication year (before 2000 versus 2000 and later); (d) clozapine versus non-clozapine SGAs; (e) randomization time point (acute versus maintenance phase); (f) determination of patient stability (⩾4 weeks versus <4 weeks); (g) first- versus multi-episode patients; and (h) haloperidol equivalent comparator dose level (<5 versus ⩾5 mg and <10 versus ⩾10 mg), calculated for non-haloperidol medications using established conversion factors.15 We also assessed the generalizability of the primary outcome results, using alternative ways to calculate relapse, that is, using preferentially raw over estimated relapse rates, and using only raw or estimated rates instead of preferring estimated relapse rate. Finally, to test whether the results could be reversed in favor of FGAs, we performed a ‘best case scenario’ analysis for FGAs. We pooled all studies where the RR for study defined relapse was ≥1.0, or >1.0, that is, we excluded all studies where SGAs had any effect (significant or non-significant) that was larger than for FGAs.
All data were entered into a funnel graph (trial effect against trial size) to investigate the likelihood of overt publication bias.16 Data were double-entered (T Kishimoto, V Agarwal) into Review Manager 5.0.25, a program developed by the Cochrane Collaboration for systematic reviews.
Search and study characteristic
We included 18 publications of 23 randomized, active drug-controlled studies with 4504 participants (Supplementary Figure 1).
The number of participants per study ranged from 32 to 690 (median: 147) and mean maximum study duration was 61.9±22.4 (range: 40–104) weeks (Table 1). There were six studies with first-episode and 17 with multiple-episode patients. Five studies were open-label, 17 were double-blind and one study was rater-masked.17 The number of studies with each individual SGA was as follows: amisulpride=3, aripiprazole=2, clozapine=4, iloperidone=3, olanzapine=6, quetiapine=1, risperidone=6, sertindole=1, and ziprasidone=1. Haloperidol was the comparator in 21/23 studies; one study used chlorpromazine;26 and one used mixed FGAs.18 Mean haloperidol equivalent dose was 11.6±8.3 mg day−1 (range: 2.9–28.5). Eighteen studies (78.3%) randomized patients in the acute phase and only five studies (21.7%) randomized patients in the maintenance phase. Eight studies (34.8%) determined patients’ stability for ⩾4 weeks and 15 studies (65.2%) determined patients’ stability <4 weeks or cross-sectionally.
Relapse definitions varied. In nine studies, relapse was not defined. In four of these,10, 18, 20, 26 we used hospitalization rate. In the remaining five, we used ‘failure to maintain response’,25 ‘psychotic exacerbation’,27 ‘failed to maintain improvement’21 and ‘dropout owing to decompensation’.17 In another study, very strict, pre-defined relapse criteria resulted in no relapse in either group.31 Therefore, the authors employed ‘marked clinical deterioration’ post hoc, which we also used.
Endpoint relapse rate
Two single studies of SGAs yielded significant superiority over FGA. These included sertindole (n=203, RR=0.29, CI: 0.10–0.84, P=0.02) and ziprasidone (n=66, RR=0.35, CI: 0.16–0.79, P=0.01). When requiring ⩾3 trials per individual antipsychotic, neither risperidone (n=1124, RR=0.75, CI: 0.56–1.00, P=0.05, I2=55%), clozapine (n=355, RR=0.72, CI: 0.47–1.10, P=0.12 I2=0%) or olanzapine (n=1140, RR=0.88, CI: 0.70–1.10, P=0.27, I2=22%) was statistically superior to FGAs in preventing relapse (Figure 1). However, when grouped together, SGAs were significantly superior to FGAs without significant heterogeneity (N=19, n=4206, 29.0 versus 37.5%, RR=0.80, CI: 0.70–0.91, P=0.0007, I2=37%; NNT=17, CI: 10–50, P=0.003) (Figure 1).
Relapse rate at 3, 6 and 12 months
Several individual SGAs were associated with significantly lower relapse rates at specific time points. This included clozapine at 3 months (P=0.03), 6 months (P=0.006), olanzapine at 6 months (P=0.0003) and sertindole (P=0.02) as well as ziprasidone (P=0.01) at 12 months. Requiring ⩾3 analyzable trials, only risperidone showed significant superiority over FGAs at both 6 months (P=0.004) and 12 months (P<0.0001) (Supplementary Figures 2–4). Pooled SGAs, however, were superior to FGAs at all pre-specified time points, that is, 3 months: 13.8 versus 17.4%, P=0.04; 6 months: 21.0 versus 28.1%, P<0.0001; 12 months: 31.4 versus 37.1%, P=0.0001.
Treatment failure; hospitalization; and dropout owing to any cause, non-adherence and intolerability
Individually, only olanzapine was superior to FGAs (P=0.03) regarding treatment failure defined as relapse and/or all-cause discontinuation, but pooled together, SGAs significantly outperformed FGAs (P=0.003) (Figure 2). Except for single-study superiority of sertindole and ziprasidone (P=0.03 each), none of the individual SGAs was superior to FGAs in preventing hospitalization (Figure 3). However, pooled together, SGAs were superior to FGAs (12.1 versus 16.9%, P=0.004).
Dropout rates for reasons other than relapse varied widely from 9.1 to 68.2% (median: 34%, 13 studies with data). Except for single study-based lower dropout for non-adherence with sertindole (P=0.02), no significant superiority was found for any individual SGA for dropout owing to any reason, non-adherence or intolerability. Even when pooled together, SGAs had only trend-level superiority over FGAs regarding dropout owing to any cause (P=0.06) (Supplementary Figure 5), non-adherence (fewer data points were available, P=0.20) (Supplementary Figure 6) and intolerability (P=0.05) (Supplementary Figure 7).
Sensitivity and subgroup analyses
The superiority of grouped SGAs regarding prevention of relapse remained significant in blinded studies (N=18, n=3519, P=0.003), pharmaceutical company-sponsored studies (N=15, n=3250, P<0.00001), studies published before and after 2000 (N=8, n=1282, P=0.0002, and N=14, n=2774, P=0.03, respectively), non-clozapine SGA studies (N=18, n=3701, P=0.002), both randomization time points (acute phase: N=17, n=3326, P=0.001; maintenance phase: N=5, n=730, P<0.00001), studies with <4 weeks or cross-sectionally assessed stability (N=14, n=2454, P=0.0006), and in first- and multi-episode patients (N=6, n=1207, P=0.02, and N=16, n=2849, P=0.0009, respectively) (Table 2). Results remained significant regardless of the haloperidol comparator dose (<5 mg/day, N=7, N=1187, P=0.01; ⩾5 mg/day, N=15, n=2869, P=0.003; <10 mg/day, N=10, n=1963, P=0.007; ⩾10 mg/day, N=12, n=2093, P=0.006). Academia-sponsored studies (N=6, n=767, P=0.05) and studies requiring validated patient stability for ⩾4 weeks showed trend-level superiority of SGAs (N=8, n=1602, P=0.08) (Table 2).
SGAs remained significantly superior over FGAs independent of whether raw relapse rates were used preferentially over estimated relapse rates (P=0.005), and whether only estimated rates or raw rates were used (P=0.0003 and P=0.02, respectively) (Table 2). Finally, performing a ‘best case scenario’ analysis for FGAs, we pooled all studies where the RR for study-defined relapse was ≥1.0. In this subsample, SGAs were not inferior to FGAs (N=9, n=836, RR=1.08, CI: 0.97–1.35, P=0.50, I2=0%). The same was true when removing the two studies with an RR=1.0, that is, when analyzing only studies that had an RR>1.0 that disfavored SGAs (N=7, n=719, RR=1.11, CI: 0.96–1.44, P=0.46, I2=0%).
Changes in psychopathology and side effects could not be formally meta-analyzed, as most studies did not provide these data separately for the stabilized subgroup in which relapse was examined.
The symmetrical funnel plot did not suggest overt publication bias (Supplementary Figure 8).
This is the largest meta-analysis to date directly comparing relapse rates in schizophrenia patients treated with SGAs or FGAs followed for ⩾6 months. We found that, whereas in some single studies individual SGAs were associated with significantly lower relapse rates and isolated other superiority regarding secondary outcomes, this was no longer the case when requiring at least three studies providing data for meta-analysis of individual drug effects. The exception was risperidone, which showed significant superiority over FGAs at both 6 months (P=0.004) and 12 months (P<0.0001) when requiring ⩾3 analyzable trials.
Of note, however, there was no instance where individual FGAs were superior to individual SGAs, either at a trial level or compared with all trials with a specific SGA. Moreover, when grouped together, SGAs as a group were superior to FGAs. Although the NNT of 17 is modest, the results were bolstered in that they were confirmed in a number of relevant sensitivity and subgroup analyses, and also extended to overall treatment failure and hospitalization, the latter of which is known to be less sensitive than relapse.34 Therefore, we consider these findings are relevant when choosing long-term treatments in clinical practice. Although SGAs were not significantly superior to FGAs regarding all-cause discontinuation, discontinuation for intolerability or non-adherence, results trended in favor of SGAs (P=0.05–0.20), and the analyzable samples for these outcomes included only 18–50% of all patients. However, these results need to be considered in the context of the cost-effectiveness discussion regarding SGAs versus FGAs.9, 35, 36, 37
Our report included 23 studies, involving 4504 participants. This extends the similar findings from the earlier meta-analysis11 that included 11 studies with 2032 patients. The inclusion criteria and methodology were similar, except that we preferred survival curve-estimated relapse rates over raw rates as our primary outcome, which we believe is a better measure as the shorter follow-up durations often found with FGAs can bias the results against SGAs, which often have a longer follow-up and observation duration during which relapse can occur. Actually, in both the prior and current meta-analyses, differences were smaller when raw relapse rates were used, but the results were not affected by the methods used to calculate relapse. Furthermore, compared with the prior meta-analysis,11 we were able to include four additional SGAs in our analyses, that is, aripiprazole, iloperidone, quetiapine and ziprasidone; we included six first episode studies; and we were able to extend the analyses by investigating multiple secondary outcomes and conducting previously unavailable sensitivity analyses that confirmed and extended the primary results. This included superiority of SGAs compared with FGAs dosed below 5 mg/day (haloperidol equivalents), whereas the comparatively high haloperidol doses used in the earlier studies had been a major shortcoming in the previously available data base.
Nevertheless, relapse rates were substantially different between prior and current analyses, even when taking into account that we preferentially used survival analyses-based rates. In the prior analysis,11 relapse rates at 1 year were 15 versus 23% for SGAs and FGAs compared with 31.4 versus 37.1% in our analysis. It appears that the low threshold definition of relapse in some more recent, large trials accounts for this difference, but SGAs demonstrated superiority regardless of whether study-defined relapse, treatment failure or hospitalization was used.
There has been much recent debate about the relative merits of SGAs over FGAs.8, 9, 10, 36, 37, 38, 39, 40 Increasingly, the heterogeneity of SGAs and FGAs with need for individualization of treatment is being stressed.8, 35, 36, 37, 38, 39 Nevertheless, different drug classes are usually determined by distinctly different mechanisms, and SGAs and FGAs differ regarding potentially relevant receptor-binding profiles. Moreover, the grouping has some historical relevance because of previous reviews and clinical trials of efficacy, effectiveness, relapse, EPS and tardive dyskinesia. Although we think the strict dichotomy has outlived its usefulness, we now have a larger series of studies comparing SGAs with haloperidol (once the leading drug worldwide), suggesting that there are modest differences regarding relapse prevention, a prevailing long-term goal in schizophrenia, regardless of high, medium or low haloperidol comparator dose, and, possibly, dropout owing to intolerability and non-adherence. The fact that we found relatively consistent differences favoring SGAs, although modest, also has heuristic implications in that some patients relapse despite adequate dopamine antagonism provided by FGAs and SGAs. Therefore, an understanding of what other mechanisms might be relevant for relapse (even in a subset of patients) is important.
However, the results of this study have to be interpreted in the context of several limitations. The data base, although larger than in the previous meta-analysis, is still limited, especially regarding individual SGAs as well as FGA comparators other than haloperidol. This limitation does not allow for a conclusive comparison of individual SGAs, which needs to be addressed by conducting additional studies. Another important limitation is the inconsistent definition of relapse. As noted, we used each study-defined relapse measure, and if no definition of relapse was available, or if the study-defined relapse criteria were considered inappropriate, we used what we judged to be the most appropriate relapse-related outcome, that is, predominantly psychiatric hospitalization. The problem of heterogeneously defined relapse is not surprising, as there is no universally accepted definition. On the other hand, this heterogeneity and broad-based definition of the primary outcome could also serve to enhance the generalizability of the results.
A further limitation is the methodological variability of the studies. For example, in many trials randomization occurred in the acute phase. To deal with this problem, we only used the subpopulation of patients who were judged to be responders or who were stable enough to be discharged, so that this subpopulation could be considered ‘at risk’ for relapse, having demonstrated clear improvement as well as subsequent, clear exacerbation from that new baseline state. The concern is that by including studies, which randomized acutely exacerbated patients, we would include only patients at risk for relapse who had responded to that specific medication for acute treatment. This could lead to a selection bias toward patients who experienced less side effects or experienced more improvement on the allocated medication. If we were limited to studies randomizing patients in the maintenance phase, only five studies would have been eligible for this meta-analysis. Furthermore, we also wanted to use the same inclusion criteria as in the previous meta-analysis.11 Moreover, this apparent bias applied to both SGAs and FGAs, and mirrors clinical practice, in that maintenance treatment is used in patients who tolerate a given treatment and who do reasonably well on it.
Another limitation is the paucity of available data on potentially relevant factors, such as EPS and adherence, as well as the use of the high-potency FGA haloperidol in 21/23 studies, which precluded subgroup analyses for mid- or low-potency FGAs. The possibility that higher EPS rates could contribute to the higher rate of relapse, either directly or indirectly through non-adherence, should be considered. For example, the possibility exists that akathisia or severe akinesia might have mimicked or contributed to apparent psychotic exacerbation. However, it is unlikely that in a maintenance study involving relatively stable patients, there would be a sufficiently sudden or dramatic increase in EPS to trigger a clinical or rating-scale threshold of relapse. This is of particular importance because haloperidol was a comparator in most studies and therefore might have facilitated an apparent advantage for SGAs owing to its higher EPS risk. However, SGAs were superior regardless of the haloperidol comparator dose. There were insufficient data to conduct a meta-analysis on EPS or adherence, but in the studies with data,17, 19, 20, 23, 31, 32 SGAs showed either significant or trend-level superiority in some of the EPS-related outcomes. Some studies provided non-adherence rates and some provided data on discontinuation for non-adherence, but these measures were generally crude. Nevertheless, we did not find significantly more non-adherence with FGAs in the nine studies with relevant data. In addition, other studies have not consistently shown that adherence with SGAs is sufficiently superior to explain the differences in relapse rates observed in our meta-analysis.41, 42
Assuming that we have identified a true difference between SGAs and FGAs for relapse prevention, unconfounded with differences in adherence or EPS, possible explanations for this finding deserve consideration. Differences in receptor-biding profiles might have some role in relapse prevention. Clearly dopamine receptor antagonism alone is insufficient to prevent all relapses as evidenced by the roughly 20% of patients who relapse within a year on long-acting injectable antipsychotics.43 It is possible that SGAs are associated with less dopamine receptor upregulation (as evidenced by lower rates of tardive dyskinesia4) and that this might also impact rates of psychotic relapse. At the same time, SGAs have outperformed FGAs on measures of subjective well-being and quality of life, raising the possibility that these might also be mediating factors in relapse risk.44 However, a detailed discussion of these possibilities is beyond the scope this report.
Finally, we acknowledge that other long-term costs of FGA and SGA treatment, such as the risk for tardive dyskinesia4, 45 and for cardiovascular adverse effects,6, 46, 47, 48, 49 require careful consideration in the individualized choice of treatments for schizophrenia patients.
Thus, although the results might appear somewhat confusing in that only several individual SGAs separated from the FGA comparator, whereas in pooled analyses SGAs were clearly superior to FGAs, we believe that our results indicate that this disconnect is likely due to a lack of power. This interpretation of the results is based on the following: First, despite inclusion of heterogeneous SGAs and study populations, there is no evidence that FGAs are superior to SGAs. Even when we restricted the analyses removing all individual studies that showed results in the direction of favoring SGAs (that is, best case scenario for FGAs in that all RRs were ≥1.0 or >1.0), the P-value for the comparison of these 9 and 7 studies was 0.50 and 0.43, respectively. Second, the superiority of combined SGAs versus FGAs was widespread, generalizing to almost all examined efficacy outcomes.
In conclusion, results from this meta-analysis suggest that, whereas individually SGAs were not consistently superior to FGAs, as a group, SGAs were associated with less study-defined relapse, overall treatment failure and hospitalization than FGAs, having a modest but clinically relevant effect size. Future relapse prevention studies should carefully assess EPS and adherence. Moreover, additional studies with a variety of SGAs using non-haloperidol FGA comparators at low-medium doses that do not produce significantly greater EPS than SGAs47 are needed to extend these findings. In particular, sufficiently large data sets are needed to allow examination of the relative merits of individual SGAs, and to guide an individualized and evidence-based maintenance treatment selection in schizophrenia.
Lieberman JA . Atypical antipsychotic drugs as a first-line treatment of schizophrenia: a rationale and hypothesis. J Clin Psychiatry 1996; 57 (Suppl 11): 68–71.
Alexander GC, Gallagher SA, Mascola A, Moloney RM, Stafford RS . Increasing off-label use of antipsychotic medications in the United States, 1995–2008. Pharmacoepidemiol Drug Safety 2011; 20: 177–184.
Rummel-Kluge C, Komossa K, Schwarz S, Hunger H, Schmid F, Kissling W et al. Second-generation antipsychotic drugs and extrapyramidal side effects: A systematic review and meta-analysis of head-to-head comparisons. Schizophr Bull. Advance online publication, 31 May 2010 (e-pub ahead of print).
Correll CU, Leucht S, Kane JM . Lower risk for tardive dyskinesia associated with second-generation antipsychotics: a systematic review of 1-year studies. Am J Psychiatry 2004; 161: 414–425.
American Diabetes Association; American Psychiatric Association; American Association of Clinical Endocrinologists; North American Association for the Study of Obesity. Consensus development conference on antipsychotic drugs and obesity and diabetes. J Clin Psychiatry 2004; 65: 267–272.
Correll CU, Manu P, Olshanskiy V, Napolitano B, Kane JM, Malhotra AK . Cardiometabolic risk of second-generation antipsychotic medications during first-time use in children and adolescents. JAMA 2009; 302: 1765–1773.
Rosenheck R, Perlick D, Bingham S, Liu-Mares W, Collins J, Warren S et al. Effectiveness and cost of olanzapine and haloperidol in the treatment of schizophrenia: a randomized controlled trial. JAMA 2003; 290: 2693–2702.
Lieberman JA, Stroup TS, Mcevoy JP, Swartz MS, Rosenheck RA, Perkins DO et al. Effectiveness of antipsychotic drugs in patients with chronic schizophrenia. N Engl J Med 2005; 353: 1209–1223.
Jones PB, Barnes TR, Davies L, Dunn G, Lloyd H, Hayhurst KP et al. Randomized controlled trial of the effect on quality of life of second- vs first-generation antipsychotic drugs in schizophrenia: Cost Utility of the Latest Antipsychotic Drugs in Schizophrenia Study (CUtLASS 1). Arch Gen Psychiatry 2006; 63: 1079–1087.
Kahn RS, Fleischhacker WW, Boter H, Davidson M, Vergouwe Y, Keet IP et al. Effectiveness of antipsychotic drugs in first-episode schizophrenia and schizophreniform disorder: an open randomised clinical trial. Lancet 2008; 371: 1085–1097.
Leucht S, Barnes TR, Kissling W, Engel RR, Correll C, Kane JM . Relapse prevention in schizophrenia with new-generation antipsychotics: a systematic review and exploratory meta-analysis of randomized, controlled trials. Am J Psychiatry 2003; 160: 1209–1222.
Tran PV, Dellva MA, Tollefson GD, Wentley AL, Beasley Jr CM . Oral olanzapine versus oral haloperidol in the maintenance treatment of schizophrenia and related psychoses. Br J Psychiatry 1998; 172: 499–505.
DerSimonian R, Laird N . Meta-analysis in clinical trials. Control Clin Trials 1986; 7: 177–188.
Higgins JP, Thompson SG, Deeks JJ, Altman DG . Measuring inconsistency in meta-analyses. BMJ 2003; 327: 557–560.
Lehman AF, Lieberman JA, Dixon LB, McGlashan TH, Miller AL, Perkins DO et al. Practice guideline for the treatment of patients with schizophrenia, second edition. Am J Psychiatry 2004; 161 (2 Suppl): 1–56.
Egger M, Davey SG, Schneider M, Minder C . Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315: 629–634.
Tamminga CA, Thaker GK, Moran M, Kakigi T, Gao XM . Clozapine in tardive dyskinesia: observations from human and animal model studies. J Clin Psychiatry 1994; 55 (Suppl B): 102–106.
Essock SM, Hargreaves WA, Covell NH, Goethe J . Clozapine's effectiveness for patients in state hospitals: results from a randomized trial. Psychopharmacol Bull 1996; 32: 683–697.
Speller JC, Barnes TR, Curson DA, Pantelis C, Alberts JL . One-year, low-dose neuroleptic study of in-patients with chronic schizophrenia characterised by persistent negative symptoms. Amisulpride v. haloperidol. Br J Psychiatry 1997; 171: 564–568.
Daniel DG, Wozniak P, Mack RJ, McCarthy BG . Long-term efficacy and safety comparison of sertindole and haloperidol in the treatment of schizophrenia. The Sertindole Study Group. Psychopharmacol Bull 1998; 34: 61–69.
Rosenheck R, Evans D, Herz L, Cramer J, Xu W, Thomas J et al. How long to wait for a response to clozapine: a comparison of time course of response to clozapine and conventional antipsychotic medication in refractory schizophrenia. Schizophr Bull 1999; 25: 709–719.
Colonna L, Saleem P, Dondey-Nouvel L, Rein W . Long-term safety and efficacy of amisulpride in subchronic or chronic schizophrenia. Amisulpride Study Group. Int Clin Psychopharmacol 2000; 15: 13–22.
Csernansky JG, Mahmoud R, Brenner R . A comparison of risperidone and haloperidol for the prevention of relapse in patients with schizophrenia. N Engl J Med 2002; 346: 16–22.
de Sena EP, Santos-Jesus R, Miranda-Scippa A, Quarantini LC, Oliveira IR . Relapse in patients with schizophrenia: a comparison between risperidone and haloperidol. Rev Bras Psiquiatr 2003; 25: 220–223.
Kasper S, Lerman MN, McQuade RD, Saha A, Carson WH, Ali M et al. Efficacy and safety of aripiprazole vs haloperidol for long-term maintenance treatment following acute relapse of schizophrenia. Int J Neuropsychopharmacol 2003; 6: 325–337.
Lieberman JA, Phillips M, Gu H, Stroup S, Zhang P, Kong L et al. Atypical and conventional antipsychotic drugs in treatment-naive first-episode schizophrenia: a 52-week randomized trial of clozapine vs chlorpromazine. Neuropsychopharmacology 2003; 28: 995–1003.
Marder SR, Glynn SM, Wirshing WC, Wirshing DA, Ross D, Widmark C et al. Maintenance treatment of schizophrenia with risperidone or haloperidol: 2-year outcomes. Am J Psychiatry 2003; 160: 1405–1412.
Schooler N, Rabinowitz J, Davidson M, Emsley R, Harvey PD, Kopala L et al. Risperidone and haloperidol in first-episode psychosis: a long-term randomized trial. Am J Psychiatry 2005; 162: 947–953.
Lieberman JA, Tollefson G, Tohen M, Green AI, Gur RE, Kahn R et al. Comparative efficacy and safety of atypical and conventional antipsychotic drugs in first-episode psychosis: a randomized, double-blind trial of olanzapine versus haloperidol. Am J Psychiatry 2003; 160: 1396–1404.
Green AI, Lieberman JA, Hamer RM, Glick ID, Gur RE, Kahn RS et al. Olanzapine and haloperidol in first episode psychosis: two-year data. Schizophr Res 2006; 86: 234–243.
Gaebel W, Riesbeck M, Wolwer W, Klimke A, Eickhoff M, von Wilmsdorff M et al. Maintenance treatment with risperidone or low-dose haloperidol in first-episode schizophrenia: 1-year results of a randomized controlled trial within the German Research Network on Schizophrenia. J Clin Psychiatry 2007; 68: 1763–1774.
Kane JM, Lauriello J, Laska E, Di Marino M, Wolfgang CD . Long-term efficacy and safety of iloperidone: results from 3 clinical trials for the treatment of schizophrenia. J Clin Psychopharmacol 2008; 28 (2 Suppl 1): S29–S35.
Crespo-Facorro B, Perez-Iglesias R, Mata I, Caseiro O, Martinez-Garcia O, Pardo G et al. Relapse prevention and remission attainment in first-episode non-affective psychosis. A randomized, controlled 1-year follow-up comparison of haloperidol, risperidone and olanzapine. J Psychiatr Res 2010; 45: 763–769.
Gitlin M, Nuechterlein K, Subotnik KL, Ventura J, Mintz J, Fogelson DL et al. Clinical outcome following neuroleptic discontinuation in patients with remitted recent-onset schizophrenia. Am J Psychiatry 2001; 158: 1835–1842.
Rosenheck RA, Leslie DL, Sindelar J, Miller EA, Lin H, Stroup TS et al. Cost-effectiveness of second-generation antipsychotics and perphenazine in a randomized trial of treatment for chronic schizophrenia. Am J Psychiatry 2006; 163: 2080–2089.
Rosenheck RA, Leslie DL, Doshi JA . Second-generation antipsychotics: cost-effectiveness, policy options, and political decision making. Psychiatr Serv 2008; 59: 515–520.
Rosenheck RA . Pharmacotherapy of first-episode schizophrenia. Lancet 2008; 371: 1048–1049.
Leucht S, Corves C, Arbter D, Engel RR, Li C, Davis JM . Second-generation versus first-generation antipsychotic drugs for schizophrenia: a meta-analysis. Lancet 2009; 373: 31–41.
Kane JM, Correll CU . Past and present progress in the pharmacologic treatment of schizophrenia. J Clin Psychiatry 2010; 71: 1115–1124.
Leucht S, Heres S, Kissling W, Davis JM . Evidence-based pharmacotherapy of schizophrenia. Int J Neuropsychopharmacol 2011; 14: 269–284.
Dolder CR, Lacro JP, Dunn LB, Jeste DV . Antipsychotic medication adherence: is there a difference between typical and atypical agents? Am J Psychiatry 2002; 159: 103–108.
Velligan DI, Weiden PJ, Sajatovic M, Scott J, Carpenter D, Ross R et al. The expert consensus guideline series: adherence problems in patients with serious and persistent mental illness. J Clin Psychiatry 2009; 70 (Suppl 4): 1–46.
Leucht C, Heres S, Kane JM, Kissling W, Davis JM, Leucht S . Oral versus depot antipsychotic drugs for schizophrenia—a critical systematic review and meta-analysis of randomised long-term trials. Schizophr Res 2011; 127: 83–92.
Lambert M, Schimmelmann BG, Schacht A, Suarez D, Haro JM, Novick D et al. Differential 3-year effects of first- versus second-generation antipsychotics on subjective well-being in schizophrenia using marginal structural models. J Clin Psychopharmacol 2011; 31: 226–230.
Rosenheck RA . Evaluating the cost-effectiveness of reduced tardive dyskinesia with second-generation antipsychotics. Br J Psychiatry 2007; 191: 238–245.
Nielsen J, Skadhede S, Correll CU . Antipsychotics associated with the development of type 2 diabetes in antipsychotic-naive schizophrenia patients. Neuropsychopharmacology 2010; 35: 1997–2004.
Rummel-Kluge C, Komossa K, Schwarz S, Hunger H, Schmid F, Lobos CA et al. Head-to-head comparisons of metabolic side effects of second generation antipsychotics in the treatment of schizophrenia: a systematic review and meta-analysis. Schizophr Res 2010; 123: 225–233.
Correll CU, Lencz T, Malhotra AK . Antipsychotic drugs and obesity. Trends Mol Med 2011; 17: 97–107.
De Hert M, Detraux J, van Winkel R, Yu W, Correll CU . Metabolic and cardiovascular adverse effects associated with antipsychotic drugs. Nat Rev Endocrinol; Advance online publication, 18 October 2011; doi:10.1038/nrendo.2011.156 (e-pub ahead of print).
Leucht S, Wahlbeck K, Hamann J, Kissling W . New generation antipsychotics versus low-potency conventional antipsychotics: a systematic review and meta-analysis. Lancet 2003; 361: 1581–1589.
This work was supported in part by The Zucker Hillside Hospital Advanced Center for Intervention and Services Research for the Study of Schizophrenia (MH090590) from the National Institute of Mental Health, Bethesda, MD, USA. The sponsor had no influence on the design, data acquisition, data analysis, data interpretation or writing of the report. We thank Gennady Gelman, MD, and Allyssa Brody, BS, for help with the literature search and data abstraction. We thank the following authors and pharmaceutical companies for providing additional, unpublished data on their studies relevant for this meta-analysis: Drs Benedicto Crespo-Facorro, Eduardo Pondé de Sena, Jeffrey A Lieberman, Robert Hamer, Eli Lilly (Drs Bruce Kinon and Virginia Stauffer), Janssen-Cilag Brazil (Dr de Sena) and Novartis Pharmaceuticals Corporation (Dr Marla Hochfeld).
Dr Kishimoto has received speaker's honoraria from Banyu, Eli Lilly, Dainippon Sumitomo, Janssen, Otsuka and Pfizer. He has received grant support from the Byoutaitaisyakenkyukai Fellowship (Fellowship of Astellas Foundation of Research on Metabolic Disorders) and Eli Lilly Fellowship for Clinical Psychopharmacology. Dr Agarwal has nothing to disclose. Dr Kishi has received speaker's honoraria from Astellas, Dainippon Sumitomo, Eli Lilly, GlaxoSmithKline, Janssen, Otsuka and Pfizer, and has received grant support from the Fellowship of Japan Research Foundation for Clinical Pharmacology. Dr Leucht received speaker/consultancy/advisory board honoraria from SanofiAventis, BMS, Eli Lilly, Essex Pharma, Astra-Zeneca, Alkermes, GlaxoSmithKline, Janssen/Johnson & Johnson, Lundbeck, Medavante and Pfizer, SanofiAventis and Eli Lilly supported research projects by SL. Dr Kane has been a consultant to Astra-Zeneca, Janssen, Pfizer, Eli Lilly, Bristol-Myers Squibb, Dainippon Sumitomo/Sepracor/Sunovion, Johnson & Johnson, Otsuka, Vanda, Proteus, Takeda, Targacept, IntraCellular Therapies, Merck, Lundbeck, Novartis, Roche, Rules Based Medicine and Sunovion, and has received honoraria for lectures from Otsuka, Eli Lilly, Esai, Boehringer-Ingelheim, Bristol-Myers Squibb and Janssen. He has received grant support from The National Institute of Mental Health. Dr Correll has been a consultant and/or advisor to or has received honoraria from Actelion, Astra-Zeneca, Bristol-Myers Squibb, Cephalon, Eli Lilly, GSK, IntraCellular Therapies, Ortho-McNeill/Janssen/J&J, Merck, Novartis, Otsuka, Pfizer and Sunovion. He has received grant support from the Feinstein Institute for Medical Research, the National Institute of Mental Health (NIMH), and the National Alliance for Research in Schizophrenia and Depression (NARSAD), BMS, Otsuka and Ortho-McNeill/Janssen/J&J.
Part of the data was presented in poster format at the 51st Annual Meeting of the New Clinical Drug Evaluation Unit (NCDEU), Boca Raton, FL, USA, 15 June 2011.
Supplementary Information accompanies the paper on the Molecular Psychiatry website
About this article
Cite this article
Kishimoto, T., Agarwal, V., Kishi, T. et al. Relapse prevention in schizophrenia: a systematic review and meta-analysis of second-generation antipsychotics versus first-generation antipsychotics. Mol Psychiatry 18, 53–66 (2013). https://doi.org/10.1038/mp.2011.143
- long-term treatment
- relapse prevention
Translational Psychiatry (2021)
Abnormally low prolactin levels in schizophrenia patients after switching to aripiprazole in a randomized trial: a biomarker for rebound in psychotic symptoms?
BMC Psychiatry (2020)
Efficacy and safety of once-monthly Risperidone ISM® in schizophrenic patients with an acute exacerbation
npj Schizophrenia (2020)
Der Nervenarzt (2020)
Long-Acting Injections in Schizophrenia: a 3-Year Update on Randomized Controlled Trials Published January 2016–March 2019
Current Psychiatry Reports (2019)