Antipsychotic medications are the first-line treatment for schizophrenia, as established by numerous randomised controlled trials (RCTs). Reference Huhn, Nikolakopoulou, Schneider-Thoma, Krause, Samara and Peter1 Despite this, more than 43% of individuals fail to exhibit significant symptom improvement and over 67% do not achieve remission. Reference Samara, Nikolakopoulou, Salanti and Leucht2 Clozapine is the gold standard for treatment-resistant schizophrenia, Reference Barnes, Drake, Paton, Cooper, Deakin and Ferrier3–Reference Samara, Lappas, Pinioti, Glarou, Fober and Christogiannis5 with a demonstrated efficacy that is higher when baseline severity is greater. Reference Samara, Lappas, Pinioti, Glarou, Fober and Christogiannis5 Its efficacy also extends beyond refractory cases to treatment-responsive schizophrenia. Reference Huhn, Nikolakopoulou, Schneider-Thoma, Krause, Samara and Peter1 Despite its clinical benefits, clozapine is underutilised, primarily due to concerns about its serious adverse effects. Reference Masuda, Misawa, Takase, Kane and Correll6,Reference Warnez and Alessi-Severini7 Recent meta-analyses have further questioned its superiority over other second-generation antipsychotics, Reference Dong, Schneider-Thoma, Bighelli, Siafis, Wang and Burschinski8–Reference Samara, Dold, Gianatsi, Nikolakopoulou, Helfer and Salanti10 suggesting that both safety concerns and uncertainties about its comparative efficacy may contribute to its limited use in clinical practice.
Current Food and Drug Administration (FDA) warnings for clozapine emphasise several serious adverse effects, including severe neutropenia, seizures, orthostatic hypotension/bradycardia/syncope, myocarditis/cardiomyopathy and an increased mortality in patients with dementia-related psychosis. 11
Beyond these boxed warnings, clozapine has been associated with other adverse effects including metabolic abnormalities, such as metabolic syndrome, dyslipidaemia, hyperglycaemia and weight gain, Reference Yuen, Kim, Procyshyn, Panenka, Honer and Barr12 as well as gastrointestinal effects – particularly impaired intestinal motility Reference Every-Palmer, Nowitz, Stanley, Grant, Huthwaite and Dunn13 that may progress to life-threatening obstruction. Reference Every-Palmer and Ellis14 Furthermore, emerging evidence suggests that clozapine may increase infection risk, most notably for pneumonia. Reference de Leon, Sanz, Norén and De las Cuevas15,Reference de Leon, Ruan, Verdoux and Wang16
While existing systematic reviews have examined clozapine’s adverse effects, their scope has been constrained by focusing on specific subpopulations or isolated side-effects. Reference Huhn, Nikolakopoulou, Schneider-Thoma, Krause, Samara and Peter1,Reference Masuda, Misawa, Takase, Kane and Correll6,Reference Dong, Schneider-Thoma, Bighelli, Siafis, Wang and Burschinski8–Reference Samara, Dold, Gianatsi, Nikolakopoulou, Helfer and Salanti10,Reference Myles, Myles, Xia, Large, Bird and Galletly17,Reference Vickers, Ramineni, Malacova, Eriksson, McMahon and Moudgil18 Moreover, recent meta-analyses have challenged the presumed unique association between clozapine and neutropenia, demonstrating comparable risks across antipsychotic classes in controlled trials, Reference Myles, Myles, Xia, Large, Bird and Galletly17 raising questions about current monitoring paradigms. To address these limitations and provide comprehensive safety guidance, we conducted the first systematic review and meta-analysis examining all clozapine-associated adverse effects across RCTs in schizophrenia. Through meta-analysis of this robust data-set, we provide incidence estimates and comparative risk profiles versus other antipsychotics to guide clinical decision-making and future research, enabling safer and more optimised clozapine utilisation.
Method
This systematic review and meta-analysis was conducted in line with PRISMA guidelines, Reference Moher, Liberati, Tetzlaff and Altman19,Reference Hutton, Salanti, Caldwell, Chaimani, Schmid and Cameron20 and the corresponding PRISMA 2020 checklist is presented in Supplementary Section 1 available at https://doi.org/10.1192/bjp.2025.10421. The review protocol was registered and made publicly available on Open Science Framework Preregistration (https://doi.org/10.17605/OSF.IO/KWQ34) and is included in Supplementary Section 2.
Selection criteria
We included individuals diagnosed with any form of schizophrenia or related psychoses, such as schizophreniform or schizoaffective disorders, based on operationalised criteria (e.g. DSM, International Classification of Diseases [ICD]) or a clinical diagnosis. No restrictions were applied regarding symptom severity, treatment responsiveness, gender, ethnicity, age, setting or comorbid conditions.
We included only RCTs that compared clozapine monotherapy with any other antipsychotic (including pooled comparator groups) or placebo. All dosages and formulations were included (oral and long-acting intramuscular [depot] preparations). In a post hoc analysis, we expanded our evaluation to include RCTs examining clozapine combination therapy (clozapine plus another psychotropic agent) to assess potential additive adverse effect risks compared with monotherapy.
Search strategy
We conducted a comprehensive search of the following sources, with no restrictions on language, covering all records from the inception of each database up to 10 October 2024: (a) MEDLINE (via Ovid), (b) Cochrane Central Register of Controlled Trials (CENTRAL), (c) Embase, (d) PsycINFO, (e) US National Institute of Health Ongoing Trials Register (ClinicalTrials.gov) and (f) the World Health Organization International Clinical Trials Registry Platform (ICTRP; www.who.int/ictrp). The search strategy combined terms for clozapine treatment and schizophrenia (Supplementary Section 3). We reviewed the reference lists of the included studies, along with previously published reviews by our team Reference Samara, Lappas, Pinioti, Glarou, Fober and Christogiannis5,Reference Samara, Dold, Gianatsi, Nikolakopoulou, Helfer and Salanti10 and others, Reference Huhn, Nikolakopoulou, Schneider-Thoma, Krause, Samara and Peter1,Reference Masuda, Misawa, Takase, Kane and Correll6,Reference Dong, Schneider-Thoma, Bighelli, Siafis, Wang and Burschinski8,Reference Schneider-Thoma, Hamza, Chalkou, Siafis, Dong and Bighelli9,Reference Myles, Myles, Xia, Large, Bird and Galletly17,Reference Vickers, Ramineni, Malacova, Eriksson, McMahon and Moudgil18 to identify additional relevant studies.
At least two reviewers (E.P. and E.G.) independently evaluated the search results, reviewed the full-text article and assessed these against the inclusion criteria. Any conflicts were resolved through consultation with a third reviewer (A.S.L. or M.S.).
Data analysis
Data extraction was independently conducted by three reviewers (E.P., E.G. and I.F.), who recorded the information on electronic Excel forms starting 1 November 2024. Any disagreements were discussed with a senior reviewer (A.S.L. or M.S.) for resolution. Study authors were contacted when further clarifications were required.
The primary outcomes were mortality; discontinuation due to adverse effects; agranulocytosis/severe neutropenia (absolute neutrophil count [ANC] ≤500/μL); 11 seizure; myocarditis/cardiomyopathy; orthostatic hypotension/bradycardia/syncope; and ileus. The secondary outcomes were leukopenia; neutropenia; tachycardia; hypertension; thromboembolism; corrected QT (QTc) interval prolongation; constipation; nausea/vomiting; weight gain (dichotomous); weight gain (continuous); glucose and/or insulin resistance levels; total cholesterol blood levels; low-density lipoprotein cholesterol (LDL) levels; triglyceride blood levels; insomnia; sedation; hypersalivation/sialorrhoea/increased saliva; dry mouth; any urological event (e.g. enuresis, nocturia, urinary incontinence); extrapyramidal adverse effects (dichotomous); extrapyramidal adverse effects (total symptoms in validated scales, continuous); akathisia; tardive dyskinesia; dystonia; Parkinsonism; headache; infection; flu-like syndromes; fever; and hospitalisation. We extracted the following information: (a) demographics (age, gender, ethnicity); (b) diagnosis and diagnostic criteria; (c) illness duration; (d) trial duration; (e) study design processes; (f) study origin; (g) publication year; (h) sponsorship; (i) treatment dose; (j) comorbidities; (k) whether clozapine was being administered to patients for the first time; and (i) whether there was documented haematological monitoring.
Two or more reviewers (E.P., E.G. and I.F.) independently assessed the risk of bias in accordance with the methodology outlined in the Cochrane Handbook for Systematic Reviews of Interventions. Reference Higgins, Altman, Gotzsche, Juni, Moher and Oxman21 The overall risk of bias was systematically classified as ‘high’, ‘moderate’ or ‘low’. Reference Cipriani, Furukawa, Salanti, Chaimani, Atkinson and Ogawa22
We employed a random-effects model Reference DerSimonian and Laird23 for the analyses using R software version 4.4.2 for Windows (R Core Team, Vienna, Austria; see https://www.r-project.org/), because it is more conservative in determining statistical significance, although it may overemphasise smaller studies. To address this, sensitivity analyses were performed to compare results with a fixed-effects model. For dichotomous outcomes, we applied the Mantel–Haenszel method Reference Efthimiou24 to estimate risk ratios. For continuous outcomes, we calculated either weighted or standardised mean differences when scales varied, estimating missing data where necessary. All analyses followed an intention-to-treat (ITT) approach, with completers’ analyses used when ITT data were unavailable. Heterogeneity was assessed using forest plots, χ 2 and I 2 statistics, and subgroup analyses were conducted for primary outcomes. Publication bias was evaluated using contour-enhanced funnel plots, the trim-and-fill method and statistical tests for asymmetry. Reference Duval and Tweedie25–Reference Peters, Sutton, Jones, Abrams and Rushton27 We also estimated the absolute event rate of primary outcomes regression model as a generalised linear mixed model (GLMM) with logit transformation, due to the rarity of events. Reference Stijnen, Hamza and Ozdemir28
Subgroup, meta-regression and sensitivity analyses were performed for the primary outcomes. Subgroup analyses included (a) age (under 18, over 65 years); (b) clozapine treatment duration; (c) population receiving clozapine for the first time during the trial versus receiving clozapine before entering the trial; (d) presence or absence of comorbidities; and (e) specific antipsychotic comparisons. The meta-regression analyses included (a) gender (male percentage) and (b) mean clozapine dose. Sensitivity analyses, utilising a fixed-effects rather than random-effects model, as well as exclusion of the following groups of studies, were performed: (a) non-double-blind studies; (b) studies with a high risk of bias; (c) studies with imputed data; (d) sponsored studies; (e) studies conducted in less developed countries; and (f) studies lacking prespecified haematological monitoring for both the clozapine and comparator groups (agranulocytosis outcome only).
Results
Description of included monotherapy studies
The PRISMA flow diagram and a detailed summary of included studies are provided in Supplementary Fig. 1a and Supplementary Table 2, respectively. A total of 69 unique RCTs with 6281 randomised participants were included. The studies were published between 1974 and 2020, with the majority conducted in Europe and the USA (37.68% each). Most studies (67.65%) employed a double-blind design while others utilised single-blind (17.65%) or open-label designs (11.76%), with three studies Reference Herken, Kaya, Beşiroğlu, Derman and Ozkan29–Reference Nam, Yang and Lee31 lacking information on blinding. Study durations ranged from 1.43 to 496.60 weeks, with a median duration of 12 weeks. The majority of trials involved adults with a mean age of 35.01 years, while only three included children or adolescents. Reference Kumra, Frazier, Jacobsen, McKenna, Gordon and Lenane32–Reference Shaw, Sporn, Gogtay, Overman, Greenstein and Gochman34 Out of the 69 RCTs, 60 reported gender distribution with 65.38% participants being male, and four Reference Gerlach, Koppelhus, Helweg and Monrad35–Reference Erlandsen38 of these trials exclusively enrolled men. The mean clozapine dose was 360.38 mg/day, ranging from 83.4 to 649 mg/day. Most studies (76.81%) used operationalised diagnostic criteria (e.g. DSM, ICD) for schizophrenia, some (21.74%) did not specify diagnostic methods and only one Reference Shopsin, Klein, Aaronsom and Collora39 relied on clinical diagnosis. Of the 69 studies, 34 involved patients with chronic responsive schizophrenia, 29 focused on treatment-resistant schizophrenia, 5 Reference Sanz-Fuentenebro, Taboada, Palomo, Aragües, Ovejero and Del Alamo40–44 involved patients with first-episode schizophrenia and one study included individuals with psychosis of unspecified origin. Reference Van Praag, Korf and Dols45 Nine studies included patients with a concurrent diagnosed psychiatric comorbidity. Among the 69 RCTs, the following comparisons with clozapine were identified: 18 with risperidone, Reference Azorin, Spiegel, Remington, Vanelle, Péré and Giguere46–Reference Breier, Malhotra, Su, Pinals, Elman and Adler55 17 with haloperidol, Reference Ciurezu, Ionescu, Nica Udangiu, Niţurad, Oproiu and Tudorache56–Reference Itoh, Miura, Yagi, Sakurai and Ohtsuka62 16 with olanzapine, Reference Meltzer63–Reference Tollefson, Birkett, Kiesler and Wood70 13 with chlorpromazine, Reference Claghorn, Honigfeld, Abuzzahab, Wang, Steinbook and Tuason71–Reference Guirguis, Voineskos, Gray and Schlieman77 6 with antipsychotics in general Reference Brunette, Dawson, O’Keefe, Narasimhan, Noordsy and Wojcik78–Reference Krakowski, Czobor, Citrome, Bark and Cooper83 (i.e. usual care), 3 with quetiapine, Reference Kumar, Chavan, Sidana and Das84,Reference McEvoy, Lieberman, Stroup, Davis, Meltzer and Rosenheck85 3 with ziprasidone, Reference Dutta, Bhowmick, Mitra, Nath, Chatterjee and Bhattacharjee86–Reference Schnell, Koethe, Krasnianski, Gairing, Schnell and Daumann88 3 with zotepine, Reference Lin, Chiu, Chen, Liou, Wang and Chen89–Reference Schimmelmann, Moritz, Karow, Schafer, Bussopulos and Golks91 1 with amisulpride, Reference Schimmelmann, Moritz, Karow, Schafer, Bussopulos and Golks91 1 with asenapine, Reference Dutta, Bhowmick, Mitra, Nath, Chatterjee and Bhattacharjee86 2 with placebo, Reference Shopsin, Klein, Aaronsom and Collora39,Reference Pickar, Owen, Litman, Konicki, Gutierrez and Rapaport92 2 with sulpiride, Reference Herken, Kaya, Beşiroğlu, Derman and Ozkan29,Reference Wu, Zhao, Liu, Zhai, Guo and Guo43 2 with thioridazine, Reference Edwards, Cocks, Burnett, Maud, Wong and Yuen93,Reference Gerlach and Simmelsgaard94 1 with fluphenazine, Reference Pickar, Owen, Litman, Konicki, Gutierrez and Rapaport92 1 with perphenazine Reference Van Praag, Korf and Dols45 and 1 with remoxipride. Reference Klieser, Strauss and Lemmer95 Ongoing studies and studies with unusable data are provided in Supplementary Sections 7 and 8, respectively.
Primary outcomes
Mortality data were drawn from 44 studies, demonstrating no difference in risk associated with clozapine compared with other antipsychotics (Mantel–Haenszel risk ratio 1.01, 95% CI: 0.50, 2.01, N = 3976; Fig. 1), with a clozapine prevalence of 0.1% (95% CI: 0%, 1.3%; Supplementary Fig. 9) and a prevalence for other antipsychotics of 0.09% (95% CI: 0.04%, 0.18%). Similarly, discontinuations due to adverse effects, examined in 47 studies, showed no difference between clozapine and other antipsychotics (Mantel–Haenszel risk ratio 1.18, 95% CI: 0.91, 1.53, N = 4812; Fig. 2). The prevalence of discontinuation due to adverse effects was 7.2% (95% CI: 5.20%, 10.0% for clozapine; Supplementary Fig. 10) and 6.1% (95% CI: 4.70%, 7.9%) for other antipsychotics.
Forest plot for the primary outcome: mortality risk. Reference comparator: clozapine. AP, antipsychotics; CI, confidence interval; MH, Mantel–Haenszel; SGA, second-generation antipsychotics.
Forest plot for the primary outcome: discontinuations due to adverse effects. Reference comparator: clozapine. AP, antipsychotics; CI, confidence interval; MH, Mantel–Haenszel.
Data on agranulocytosis/severe neutropenia were available from 23 studies. Clozapine was associated with an increased risk of agranulocytosis compared with other antipsychotics, although this did not reach statistical significance by a small margin (Mantel–Haenszel risk ratio 2.81, 95% CI: 0.97, 8.12, N = 3247; Fig. 3). The event itself was rare: 0.7% (95% CI: 0.3%, 1.7%) of clozapine-treated patients experienced agranulocytosis (Supplementary Fig. 7) compared with 0.24% (95% CI: 0.08%, 0,72%) of those receiving other antipsychotics.
Forest plot for the primary outcome: agranulocytosis/severe neutropenia risk. Reference comparator: clozapine. AP, antipsychotics; CI, confidence interval; MH, Mantel–Haenszel.
Seizure outcomes were assessed in 15 studies. Clozapine use was associated with a significantly higher risk of seizures compared with other antipsychotics (Mantel–Haenszel risk ratio 3.61, 95% CI: 1.80, 7.25, N = 2012; Fig. 4). Seizure prevalence was 3.1% (95% CI: 1.3%, 7.2%) with clozapine (Supplementary Fig. 8) versus 0.85% (95% CI: 0.42%, 1.72%) with other antipsychotics.
Forest plot for the primary outcome: seizures risk. Reference comparator: clozapine. AP, antipsychotics; CI, confidence interval; MH, Mantel–Haenszel.
Reported cardiac adverse effects (excluding tachycardia, cardiomyopathy and myocarditis) included orthostatic hypotension, bradycardia and syncope. These were evaluated in 21 studies, which suggested an increased risk with clozapine (Mantel–Haenszel risk ratio 1.66, 95% CI: 1.00, 2.77, N = 2914; Fig. 5). Prevalence was 11% (95% CI: 7.8%, 15.1%) for clozapine (Supplementary Fig. 11) versus 6.62% (95% CI: 3.97%, 11.0%) for other antipsychotics.
Forest plot for the primary outcome: orthostatic hypotension, bradycardia and syncope risk. Reference comparator: clozapine. AP, antipsychotics; CI, confidence interval; MH, Mantel–Haenszel.
Only two studies provided data for myocarditis/cardiomyopathy, demonstrating no elevated risk associated with clozapine compared with other antipsychotics (Mantel–Haenszel risk ratio 0.33, 95% CI: 0.01, 8.13, N = 997; Supplementary Fig. 6). Only one study Reference Howanitz, Pardo, Smelson, Engelhart, Eisenstein and Losonczy96 provided data for ileus, reporting no case in the clozapine-treated group (0/24 patients) and one case in the control (chlorpromazine) group (1/18 patients).
Secondary outcomes
Across all secondary outcomes, clozapine use was associated with an elevated risk of leukopenia, hypersalivation, sedation, tachycardia, hypertension, constipation, nausea/vomiting, fever, flu-like syndrome and headache compared with other antipsychotics. No elevated risk was found for neutropenia, dry mouth, insomnia or extrapyramidal symptoms. A small to moderate increase was observed for cholesterol, triglycerides and glucose levels, while a slight reduction in the QTc interval was noted. Risk ratios, confidence intervals and heterogeneity metrics for the outcomes above, along with the corresponding forest plots, are provided in Supplementary Section 12.
Subgroup and sensitivity analysis
Subgroup, meta-regression and sensitivity analyses did not show any significant impact on the overall results. No significant relationship was found between dose, duration or other factors and the analysed outcomes (Supplementary Sections 13 and 14).
Risk of bias
Out of the 69 studies, 8 (11.59%) were considered as having a low overall risk of bias, 29 (42.03%) were considered as having a moderate overall risk while 32 (46.38%) were considered as having a high overall risk. The risk of bias summary plot and assessment for each individual study are provided in Supplementary Figs 3 and 4, respectively.
Publication bias
Publication bias was evaluated via visual inspection of contour-enhanced funnel plots, Egger’s regression test and the trim-and-fill method (Supplementary Sections 15.1., 15.2. and 15.3). The funnel plots showed no substantial asymmetry, and Egger’s test indicated no significant small-study effects. Collectively, these findings suggest that publication bias is not a significant concern in this meta-analysis.
Post hoc analysis of studies on clozapine combinations
A total of 47 RCTs provided usable data: 20 on combination with another antipsychotic, 5 on combination with an antidepressant, 6 on combination with an antiepileptic and 16 with other drugs. The PRISMA flow diagram (Supplementary Fig. 1b) and details of the included studies (Supplementary Table 3) are provided in the supplementary file. The trial duration varied from 3 to 52 weeks, with a median of 12 weeks. The treatment duration of clozapine prior to trial initiation ranged from at least 3 months to over 120 months, with a median of 6 months.
Among the 46 studies, 21 provided data on agranulocytosis (N = 894), with no cases observed. Eight studies provided seizure data (N = 313), with no incidents recorded. Nine studies reported orthostatic hypotension (Mantel–Haenszel risk ratio 0.89, 95% CI: 0.38, 2.09, N = 425; Supplementary Section 170 and Supplementary Fig. 169). Mortality was assessed in 37 RCTs (N = 225), with no cases reported except for one death unrelated to clozapine treatment in a single RCT. Reference Nielsen, Emborg, Gydesen, Dybbro, Aagaard and Haderup97 Thirty-nine studies evaluated treatment discontinuation due to adverse events (Mantel–Haenszel risk ratio 1.25, 95% CI: 0.82, 2.05, N = 1965; Supplementary Fig. 169). Finally, 5 studies examined myocarditis (N = 174), with no cases observed. One study reported a single case of ileus in the clozapine monotherapy group (1/25) Reference Nielsen, Emborg, Gydesen, Dybbro, Aagaard and Haderup97 but no cases in the clozapine plus sertindole group (0/25), after a mean clozapine exposure of 7.5 years.
Discussion
To our knowledge, this is the largest and most comprehensive meta-analysis of RCTs evaluating clozapine’s safety profile in patients with schizophrenia. The clinical implications of our findings are discussed regarding mortality, haematological complications, seizure risk, cardiovascular adverse events (including myocarditis and related symptoms) and gastrointestinal issues, which represent clinically significant safety concerns.
Our meta-analysis found no difference in mortality risk between clozapine and other antipsychotics (risk ratio 1.01, 95% CI: 0.50, 2.01). Large-scale observational studies have consistently shown a protective effect of clozapine on all-cause mortality. Reference Taipale, Tanskanen, Mehtälä, Vattulainen, Correll and Tiihonen98,Reference Vermeulen, VanRooijen, VanDeKerkhof, Sutterland, Correll and De Haan99 This apparent discrepancy should be interpreted in light of the methodological differences between RCTs and observational studies. Observational data typically reflect long-term, real-world treatment trajectories; in contrast, most RCTs included in our analysis had relatively short follow-up periods (median duration 12 weeks) and were not specifically designed to assess rare outcomes, such as mortality. Furthermore, RCTs often apply strict inclusion criteria that may exclude higher-risk populations where clozapine’s potential mortality benefit could be more evident. Therefore, rather than contradicting observational findings, our results highlight the limitations of the available RCT evidence in assessing such outcomes.
Clozapine revealed an increased risk of agranulocytosis (risk ratio 2.81, 95% CI: 0.97, 8.12) compared with other antipsychotics, although the wide confidence interval indicates uncertainty regarding the exact estimate. Our analysis of 23 RCTs (N = 3247 total; 1549 clozapine-treated patients versus 1698 controls) showed an absolute agranulocytosis risk of 0.7% with clozapine – consistent with observational estimates that range between 0.4 and 1.4%. Reference Northwood, Myles, Clark, Every-Palmer, Myles and Kisely100–102 Thus, while the relative risk is elevated the absolute risk is low. The interpretation requires caution given both the rare-event nature of agranulocytosis and potential surveillance bias in clinical practice, although 67.7% of included trials used double-blind methodology with standardised monitoring, minimising differential assessment between treatment arms.
Our analysis also corroborates the fact that clozapine’s effects on neutrophils appear binary Reference Fernandez-Egea and McCutcheon103 : while it may lead to agranulocytosis, it does not increase the incidence of milder forms of neutropenia (Supplementary Fig. 14). Current monitoring protocols often capture cases of transient or mild neutropenia that may not require treatment discontinuation. For instance, transient neutropenia lasting 2–5 days and benign weekly variations in neutrophil count typically do not necessitate stopping clozapine. Reference Fernandez-Egea and McCutcheon103
In light of these, the FDA has updated its guidelines, replacing white blood cell (WBC) monitoring with ANC monitoring and lowering the threshold for treatment discontinuation from 1500/mm3 to 1000/mm3. Reference Fernandez-Egea and McCutcheon103 Emerging European data similarly suggest that current haematological monitoring protocols may overestimate clozapine’s agranulocytosis risk, Reference Oloyede, Bachmann, Dzahini, Lopez Alcaraz, Singh and Vallianatu104 prompting calls to revise and relax monitoring requirements. Reference Fernandez-Egea and McCutcheon103 These updates reflect an evolving understanding of clozapine’s risk–benefit profile and aim to enhance patient access and avoid unnecessary treatment discontinuations.
In the analysis of monotherapy trials, all RCTs involved patients initiating antipsychotic treatment – a period known to be associated with the highest risk of agranulocytosis for clozapine, and possibly for other antipsychotics as well. This risk for clozapine decreases substantially over time, becoming significantly lower after 2–3 years of continuous treatment. Reference Fernandez-Egea and McCutcheon103 Thus, long-term clozapine use, when carefully managed, may carry a lower risk than is commonly perceived. Our review of clozapine combination RCTs, which involved participants on clozapine maintenance therapy (median clozapine treatment duration at study entry, 6 months), reported no cases of agranulocytosis in either group (clozapine monotherapy versus clozapine combined with another drug).
The risk of seizures was more than tripled compared with that for other antipsychotics, consistent with findings from prior studies. Reference Wu, Wang, Yeh and Liu105,106 Monitoring plasma concentrations is crucial, especially because levels above 1000 ng/mL are associated with higher risk. Reference Hatano, Yamada, Matsuzaki, Yokoi, Saito and Yamada107 Prophylactic anti-epileptic therapy and minimisation of polypharmacy could help mitigate seizures, particularly in high-risk patients. Reference Hatano, Yamada, Matsuzaki, Yokoi, Saito and Yamada107 However, our review of clozapine combination RCTs revealed no incidence of seizure in either group, with a median clozapine dose of 397 mg/day (range 300–503) in the augmentation group and 362.2 mg/day (range 290.6–533.3) in the monotherapy group. Nevertheless, only 8 studies provided relevant data and the sample size was relatively small (N = 313), which may have affected the reliability of these findings.
Myocarditis represents a rare but clinically important adverse effect of clozapine, Reference Siskind, Sidhu, Cross, Chua, Myles and Cohen108 although its diagnosis remains challenging due to substantial overlap with common clozapine-related side-effects including tachycardia, fever and flu-like symptoms – manifestations that may also reflect the drug’s haematological complications or immunoinflammatory effects. Reference Martins, Moura, Mendes, Ribeiro, Arnaud and Gama109,Reference Verdoux, Quiles and De Leon110 While biomarkers (troponin, C-reactive protein [CRP]) and clinical assessment can support early detection, definitive diagnosis often requires cardiac imaging Reference Ronaldson, Taylor, Fitzgerald, Topliss, Elsik and McNeil111,Reference Ronaldson, Fitzgerald and McNeil112 because the gold-standard endomyocardial biopsy is rarely feasible in clinical practice. Reference Bellissima and Tingle113 Our meta-analysis found no significant difference in myocarditis risk between clozapine and other antipsychotics, although this finding requires cautious interpretation given limited data and the increased clozapine risk of fever (Mantel–Haenszel risk ratio 2.39, 95% CI: 1.50, 3.83) and tachycardia (Mantel–Haenszel risk ratio 2.10, 95% CI: 1.62, 2.74) that may mimic or obscure cardiac pathology. Diagnostic uncertainty leads to frequent misattribution Reference Knoph, Morgan, Palmer, Schak, Owen and Leloux114 – one case series found that two-thirds of suspected clozapine-induced myocarditis cases were probably misdiagnosed when reviewed retrospectively, often due to concurrent viral illnesses or insufficient diagnostic evidence. Reference Richardson, Greenway and Bousman115 Most true cases emerge within the first treatment month, Reference De Las Cuevas, Sanz, Ruan and De Leon116 highlighting a critical monitoring window where heightened clinical vigilance is warranted despite these diagnostic challenges. In light of these diagnostic complexities, a recent Delphi study 117 incorporating multidisciplinary expert input proposed refined monitoring protocols that emphasise slow clozapine titration, and systematic use of hs-troponin and CRP testing, during the high-risk initial weeks. The consensus underscored the value of cardiac imaging – preferably echocardiography or cardiac magnetic resonance imaging – for confirmation of suspected cases, aiming to reduce both under- and overdiagnosis. 117 These developments reflect a shift from reliance on case reports and expert opinion towards a more standardised, evidence-informed framework for early identification and safer rechallenge in selected patients. 117
The findings regarding gastrointestinal adverse effects revealed an increased risk of constipation (Mantel–Haenszel risk ratio 1.50, 95% CI: 1.09, 2.06), confirming findings from the existing literature. Reference Shirazi, Stubbs, Gomez, Moore, Gaughran and Flanagan118,Reference Wagner, Siafis, Fernando, Falkai, Honer and Röh119 Clozapine-induced constipation may lead to ileus, bowel obstruction, ischaemic colitis, gastrointestinal necrosis or toxic megacolon, Reference Shirazi, Stubbs, Gomez, Moore, Gaughran and Flanagan118 and the likelihood of these gastrointestinal adverse effects causing death is approximately tenfold more than that for severe neutropenia. Reference Lam and Ip120 Ileus occurs after an average of 4 years of clozapine use and is more closely linked to the maintenance phase than to the initial period of treatment. Reference Tomulescu, Uittenhove and Boukakiou121 In the present meta-analysis, only 1 case of ileus was reported after a mean exposure of 7.5 years. Reference Nielsen, Emborg, Gydesen, Dybbro, Aagaard and Haderup97
Several limitations should be considered when interpreting our findings. First, the predominance of short-duration trials (median duration 12 weeks) with a high risk of bias (particularly in selective reporting) yielded limited data for rare but critical adverse events. For example, myocarditis and ileus were reported in only a few RCTs, which limits the generalisability of these safety outcomes. Second, while differences in diagnostic characteristics (e.g. illness chronicity or treatment resistance status) could theoretically influence results, such clinical variables primarily affect efficacy outcomes and are unlikely to substantially modify clozapine’s adverse effect profile. Third, we observed substantial heterogeneity in outcome definitions and monitoring protocols across studies, which may have influenced the pooled effect estimates. Fourth, our post hoc comparison of combination therapy versus monotherapy, while clearly labelled as exploratory and analysed separately from protocol-defined outcomes, requires prospective validation. Finally, our selection of multiple primary outcomes based on regulatory safety priorities – although clinically justified – deviates from standard recommendations to limit primary outcomes in systematic reviews, and may have increased interpretive complexity.
In summary, our meta-analysis confirms clozapine’s elevated but rare risk of early-onset agranulocytosis and increased risk of seizures, constipation and other adverse effects. While these risks warrant vigilance, they must be balanced against clozapine’s unique benefits for treatment-resistant schizophrenia. Future research should clarify dose- and time-dependent risks in real-world settings to optimise clozapine’s risk–benefit profile.
Supplementary material
The supplementary material is available online at https://doi.org/10.1192/bjp.2025.10421
Data availability
The data that support the findings of this study are available on request from the corresponding author, M.S., upon reasonable request.
Acknowledgements
I.F. and B.H. are supported by the Polish Ministry of Science and Higher Education (‘The Excellence Initiative – Research University’ programme), grant no. 0320/2020/20. We thank all study authors who responded to our data requests. We also thank Prof. Dan Siskind, Dr Oliver Freudenreich, Prof. Susan Essock and Dr Selene Veerman for their responses to our requests.
Author contributions
M.S. and E.P. conceptualised and designed the study, with input from A.S.L., E.G., B.H., S.S., N.C., A.N. and S.L. E.P. and E.G. coordinated and managed the planning and execution of the research activities, and also conducted the literature screening. Data extraction was carried out by E.P., E.G. and I.F. Statistical analyses were performed by E.P. and A.N. E.P. drafted the initial version of the manuscript, with substantial contributions from M.S., A.S.L. and S.L. All authors reviewed and approved the final version of the manuscript. All authors had full access to the data and share responsibility for the decision to submit the manuscript for publication.
Declaration of interest
M.S. has received honoraria as a consultant/advisor and/or for lectures from Recordati, Lundbeck and Viatris. S.L. has received honoraria as a consultant/advisor and/or for lectures from Angelini, Böhringer Ingelheim, Geodon and Richter, Janssen, Johnson & Johnson, Lundbeck, LTS Lohmann, MSD, Otsuka, Recordati, Sanofi Aventis, Sandoz, Sunovion, TEVA, Eisai, Rovi, Medichem and Mitsubishi. All other authors declare no financial relationships with commercial interests. A.N. was supported by Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), grant no. NI 2226/1-1 and Project-ID 499552394 – SFB 1597.
Open access
eLetters
No eLetters have been published for this article.