Skip to main content
NCBI home page
World Psychiatry logoLink to World Psychiatry
. 2023 Jan 14;22(1):116–128. doi: 10.1002/wps.21036

Metabolic side effects in persons with schizophrenia during mid‐ to long‐term treatment with antipsychotics: a network meta‐analysis of randomized controlled trials

Angelika Burschinski 1, Johannes Schneider‐Thoma 1, Virginia Chiocchia 2,3, Kristina Schestag 1, Dongfang Wang 1, Spyridon Siafis 1, Irene Bighelli 1, Hui Wu 1, Wulf‐Peter Hansen 4, Josef Priller 1,5,6,7, John M Davis 8,9, Georgia Salanti 2, Stefan Leucht 1
PMCID: PMC9840505  PMID: 36640396

Abstract

Metabolic side effects of antipsychotic drugs can have serious health consequences and may increase mortality. Although persons with schizophrenia often take these drugs for a long time, their mid‐ to long‐term metabolic effects have been studied little so far. This study aimed to evaluate the mid‐ to long‐term metabolic side effects of 31 antipsychotics in persons with schizophrenia by applying a random‐effects Bayesian network meta‐analysis. We searched the Cochrane Schizophrenia Group's Study‐Based Register of Trials (up to April 27, 2020) and PubMed (up to June 14, 2021). We included published and unpublished, open and blinded randomized controlled trials with a study duration >13 weeks which compared any antipsychotic in any form of administration with another antipsychotic or with placebo in participants diagnosed with schizophrenia. The primary outcome was weight gain measured in kilograms. Secondary outcomes included “number of participants with weight gain”, fasting glucose, total cholesterol, low‐density lipoprotein cholesterol, high‐density lipoprotein cholesterol, and triglycerides. We identified 137 eligible trials (with 35,007 participants) on 31 antipsychotics, with a median follow‐up of 45 weeks. Chlorpromazine produced the most weight gain (mean difference to placebo: 5.13 kg, 95% credible interval, CrI: 1.98 to 8.30), followed by clozapine (4.21 kg, 95% CrI: 3.03 to 5.42), olanzapine (3.82 kg, 95% CrI: 3.15 to 4.50), and zotepine (3.87 kg, 95% CrI: 2.14 to 5.58). The findings did not substantially change in sensitivity and network meta‐regression analyses, although enriched design, drug company sponsorship, and the use of observed case instead of intention‐to‐treat data modified the mean difference in weight gain to some extent. Antipsychotics with more weight gain were often also among the drugs with worse outcome in fasting glucose and lipid parameters. The confidence in the evidence ranged from low to moderate. In conclusion, antipsychotic drugs differ in their propensity to induce metabolic side effects in mid‐ to long‐term treatment. Given that schizophrenia is often a chronic disorder, these findings should be given more consideration than short‐term data in drug choice.

Keywords: Antipsychotic drugs, metabolic side effects, weight gain, glucose, cholesterol, triglycerides, chlorpromazine, clozapine, olanzapine, zotepine, schizophrenia

Antipsychotic drugs are the core treatment for schizophrenia, because they are efficacious in acute episodes 1 and in preventing relapses 2 . Consequently, many persons with schizophrenia take antipsychotics for years, or even lifetime3, 4. However, antipsychotics also have considerable side effects 1 . Metabolic side effects can manifest as weight gain, changes in cholesterol and triglyceride metabolism (dyslipidaemia), and dysregulation of glucose homeostasis (insulin resistance extending to diabetes) 5 . They are associated with cardiovascular diseases, including myocardial infarction and stroke6, 7, 8. Therefore, metabolic side effects of antipsychotics are likely to contribute to the average 14.5 years reduced life‐span of individuals with schizophrenia 9 . Furthermore, weight gain is associated with decreased quality of life 10 and treatment non‐adherence11, 12, the latter resulting in poor treatment outcome and psychotic relapses.

As antipsychotic drugs do not differ much in efficacy 13 , guidelines recommend that the choice of the drug should be primarily informed by their side effects14, 15. Recently, a network meta‐analysis compared the metabolic effects of 18 antipsychotics during acute treatment of schizophrenia in studies with a median treatment duration of 6 weeks 16 . However, antipsychotics are also used for prevention of relapses, and individuals take them for much longer periods of time. Therefore, the aim of the current network meta‐analysis was to investigate the mid‐ to long‐term metabolic effects of these drugs in randomized controlled trials (RCTs). Such knowledge should be highly relevant for clinical practice and contribute to tailored drug choice.

METHODS

Inclusion criteria and search strategy

We report following the Preferred Reporting Items for System­atic Reviews and Meta‐Analyses (PRISMA) extension statement for network meta‐analyses. The study protocol was registered with PROSPERO (registration number: CRD42020175414) and published 17 .

We included mid‐term and long‐term randomized controlled antipsychotic drug trials (>3 and >6 months, respectively), following the classification of the Cochrane Schizophrenia Group 18 . Trials were included irrespective of their blinding and study setting. However, trials conducted in mainland China were ex­cluded due to raised quality concerns19, 20, 21, and trials with a ran­domization process at high risk of bias were also excluded. More­over, continuation studies in which only responders of the core trial could participate were excluded, because this corrupts randomization.

Studies were included if at least 80% of the trial participants had a diagnosis of schizophrenia, schizophreniform or schizoaffective disorder, irrespectively of the diagnostic criteria. There were no restrictions concerning participants' stage of the disease, age, gender or ethnicity.

We included all second‐generation antipsychotics (SGAs) avail­able in Europe or the US, and a selection of first‐generation anti­psychotics (FGAs) informed by a survey of international schizo­phre­nia experts 22 , administered as monotherapy – namely, ami­sul­pride, ari­piprazole, asenapine, benperidol, brexpiprazole, cari­prazine, chlorpromazine, clopenthixol, clozapine, flupentixol, flu­phenazine, flus­pirilene, haloperidol, iloperidone, levomeprom­a­zine, loxap­ine, lumateperone, lurasidone, molindone, olanzapine, paliperidone, penfluridol, perazine, perphenazine, pimozide, quetiapine, ris­peridone, sertindole, sulpiride, thioridazine, tio­tix­ene, trifluoperazine, ziprasidone, zotepine, and zuclopen­thixol.

Oral and long‐acting injectable (LAI) formulations of one compound were considered as different interventions, because their side effect profile could differ due to pharmacokinetic or adherence issues23, 24, but were combined in a post‐hoc sensitivity analysis. We included all study arms with doses within the target to maximum range according to the International Consensus Study on Antipsychotic Dosing 25 . Only for specific populations such as individuals with first episode or primarily negative symptoms, for which clinically different dosing regimens are recommended, we included lower doses.

We searched the Cochrane Schizophrenia Group's Study‐Based Register of Trials (compiled by monthly searches in multiple electronic databases and trial registries up to April 27, 2020), PubMed (last update on June 14, 2021) and related systematic reviews23, 26, 27, 28, 29, 30, 31, 32, 33 (see also supplementary information). Two reviewers (AB, DW) independently screened the searches; in case of disagreement, a third reviewer (JS‐T or SL) was involved.

Outcomes, data extraction and evaluation of study risk of bias

The primary outcome was weight gain in kilograms (kg). Secondary outcomes were the “number of participants with weight gain” (≥7% from baseline preferred to other definitions), and continuous measurements of fasting glucose, total cholesterol, low density lipoprotein (LDL) cholesterol, high density lipoprotein (HDL) cholesterol, and triglycerides. All outcomes were extracted at study endpoint.

Following our protocol, we also extracted data for infrequently reported outcomes – such as body mass index, waist circumference, hemoglobin A1c (HbA1c), homeostasis model assessment of insulin resistance (HOMA‐IR), and insulin – but did not consider them for further analysis due to scarcity of data. Additionally, study characteristics (study duration, blinding, criteria used to diagnose schizophrenia), population characteristics (baseline weight, age, gender, ethnicity, and lifetime exposure to antipsychotics – if not available, duration of illness was used as a proxy), and treatment characteristics (drug company sponsorship, antipsychotic dose) were extracted.

Two reviewers (AB, KS) extracted data for each included study in specifically customized digital forms in a Microsoft Access database and evaluated risk of bias using Cochrane's Risk of Bias 2 tool 34 . Conflicting entries were automatically detected and discussed, if needed, with a third reviewer (JS‐T or SL) or the original authors. Original authors and drug companies responsible for included studies published during the past 20 years were also contacted via e‐mail by AB and SL for missing information.

Data synthesis and evaluation of confidence in the evidence

Pairwise meta‐analyses were performed in a frequentist setting, while network meta‐analyses were performed in a Bayesian setting, both using the random effects model. We synthesized continuous outcomes with mean differences (MDs) and dichotomous outcomes using odds ratios (ORs), both presented with 95% credible intervals (CrIs).

For each outcome, we assumed a common heterogeneity variance (τ 2 ) across comparisons. The magnitude of heterogeneity was judged by comparing τ2 to its empirical distribution35, 36 and by considering the width of the prediction intervals. Statistical inconsistency was evaluated using the SIDE‐test for each comparison 37 and the design‐by‐treatment interaction test for the overall network 38 .

To assess the plausibility of the transitivity assumption, we compared the distribution of key study characteristics across studies grouped by comparison. To explore sources of heterogeneity and inconsistency, we also planned network meta‐regressions for base­line weight, age, gender, ethnicity, lifetime exposure to anti­psy­chot­ics, drug company sponsorship, and study duration.

We performed sensitivity analyses by analyzing only observed cases, and by excluding non‐double blind studies, studies with an overall assessment of high risk of bias, studies with enriched design, studies not using operationalized criteria to diagnose schizophrenia, and studies in which participants had minimal prior exposure to antipsychotics (e.g., children and first episode). We also performed a post‐hoc analysis excluding doses at the lower and upper ends of the range recommended by the International Consensus Study on Antipsychotic Dosing 25 .

To investigate the presence of small‐study effects (potentially associated with publication bias), we performed – for the primary outcome – a comparison‐adjusted funnel plot 39 and a contour‐enhanced funnel plot of all drugs versus placebo 40 .

All analyses were performed in R. We conducted Bayesian net­work meta‐analyses using the BUGSnet package 41 , and network meta‐regression analyses using self‐programmed routines with the rjags package 42 . Frequentist network and pairwise meta‐anal­yses were performed with the netmeta and meta packages43, 44. The confidence in the network meta‐analysis estimates was evaluated for the primary outcome with the Confidence in Network Meta‐Analysis (CINeMA) framework 45 .

RESULTS

Description of included studies

We identified 12,690 references. After title/abstract screening, we assessed 2,501 full‐text articles and included 2,039 reports on 349 trials (see Figure 1).

Figure 1.

Figure 1

Open in a new tab

Flow chart of the study selection process. RCT – randomized controlled trial

One hundred thirty‐seven trials with 35,007 participants and 31 different antipsychotics provided usable data. The median average age of participants was 38.9 years (interquartile range, IQR: 35.3‐41.4); the median trial duration was 45 weeks (IQR: 26‐52); the median percentage of women was 37% (IQR: 29‐43); and 70% (96 of 137) of the trials were double‐blind. We found no clear evidence of differences in study characteristics across comparisons. Although the number of studies per comparison was small, we judged that there were no clear violations of the transitivity assumption (see supplementary information).

Primary outcome: weight gain

One hundred ten trials on 28 antipsychotics (N=29,215 participants with an average baseline weight of 76.55 kg) contributed to the network meta‐analysis for the primary outcome (weight gain). The network plot is provided in Figure 2.

Figure 2.

Figure 2

Open in a new tab

Network plot of primary outcome “weight gain”. The lines link treatments that were directly compared in trials. The thickness of the lines corresponds to the number of trials evaluating the comparison. The size of the nodes corresponds to the number of participants assigned to the treatment. LAI – long‐acting injectable.

The network estimates and corresponding 95% CrI for each drug versus placebo are reported in Figure 3. Medication admin­istration is oral if not otherwise stated. Most drugs were associated with more weight gain than placebo. The following drugs produced on average more than 2 kg weight gain in excess to placebo: chlorpromazine (MD: 5.13), clozapine (MD: 4.21), olanzapine oral/LAI (MD: 3.82/3.60), zotepine (MD: 3.87), pimozide (MD: 6.16), and sertindole (MD: 2.30). The following drugs produced on average between 1 and 2 kg weight gain in excess to placebo: risperidone LAI/oral (MD: 2.00/1.87), brexpiprazole (MD: 1.91), paliperidone oral/LAI (MD: 1.73/1.43), quetiapine (MD: 1.59), and amisulpride (MD: 1.43). The following drugs produced on average less than 1 kg weight gain in excess to placebo: iloperidone (MD: 0.78), asenapine (MD: 0.73), cariprazine (MD: 0.62), perphenazine (MD: 0.61), and aripiprazole (MD: 0.41). The following drugs were similar to placebo: flupentixol (MD: 0.10), aripiprazole LAI (MD: 0.00), lurasidone (MD: –0.06), haloperidol (MD: –0.01), and ziprasidone (MD: –0.16). Three drugs produced on average a weight loss compared to placebo: fluspirilene LAI (MD: –9.13), haloperidol LAI (MD: –2.53), and fluphenazine LAI/oral (MD: –1.94/–1.30). However, their 95% CrIs were wide (i.e., the estimates were imprecise) and even include the possibility of small weight gain. MDs between drugs and results of relevant pairwise meta‐analyses are provided in the supplementary information.

Figure 3.

Figure 3

Open in a new tab

Forest plot of antipsychotic drugs vs. placebo for the primary outcome “weight gain”. Network meta‐analysis estimates of treatment effect of each drug vs. placebo are reported as mean differences (MDs) and 95% credible intervals (CrIs). The order of treatments is according to surface under the cumulative ranking curve (SUCRA) ranking. LAI – long‐acting injectable, AMI – amisulpride, ARI – aripiprazole, ASE – asenapine, BRE – brexpiprazole, CAR – cariprazine, CLO – clozapine, CPZ – chlorpromazine, FLP – fluspirilene, FLU – fluphenazine, FPX – flupentixol, HAL – haloperidol, ILO – iloperidone, LUR – lurasidone, OLA – olanzapine, PAL – paliperidone, PER – perphenazine, PIM – pimozide, PLB – placebo, QUE – quetiapine, RIS – risperidone, SER – sertindole, ZIP – ziprasidone, ZOT – zotepine.

No evidence of inconsistency was found (see supplementary information). The heterogeneity standard deviation common‐τ was 0.82 for MD and 0.15 on the standardized mean difference (SMD) scale, which can be interpreted as low to moderate when considering empirical distributions and prediction intervals (see also supplementary information).

In network meta‐regressions, we found that the MD of any antipsychotic versus placebo was on average 0.45 kg (95% CrI: 0.01 to 0.89) higher in sponsored than in non‐sponsored study arms. Adjusting for drug company sponsoring reduced common‐τ from 0.82 to 0.65. Other possible effect modifiers showed no clear effect (see also supplementary information).

In sensitivity analyses, when studies with enriched design were excluded, all antipsychotics showed larger MDs (on average 0.63 kg) compared to the main analysis; and observed cases (available for 21 antipsychotics) yielded more pronounced differences in MDs versus placebo, ranging from –10.63 to 6.42 kg (see also supplementary information).

Despite these observed effects on treatment results, the rankings remained similar in all network meta‐regressions and sensitivity analyses.

We found no clear indication of small‐study effects and publication bias. The overall risk of bias was “some concerns” for 72% (79 of 110) and “high” for 28% (31 of 110) of studies. The confidence in the network meta‐analysis estimates was low in 276, moderate in 123 and very low in 7 comparisons (see also supplementary information).

Secondary metabolic outcomes

The results for “number of participants with weight gain” were very similar to the primary outcome weight gain (see supplementary information). For lipid and glucose outcomes, less data were available for most antipsychotics, with no data for zotepine and the older antipsychotics except haloperidol and perphenazine.

Drugs associated with weight gain were often also associated with worse outcomes in fasting glucose and lipid parameters (see Table 1). The ranges of the MDs in mg/dl compared to placebo were as follows: from 7.64 (95% CrI: 3.17 to 13.20) for olanzapine LAI to –0.67 (95% CrI: –5.40 to 4.24) for ziprasidone concerning fasting glucose (see also Figure 4); from 12.65 (95% CrI: 8.73 to 16.51) for olanzapine to –4.69 (95% CrI: –10.39 to 1.23) for ziprasidone concerning total cholesterol (see also Figure 5); from 9.59 (95% CrI: 3.61 to 15.49) for olanzapine LAI to –1.92 (95% CrI: –5.64 to 1.96) for aripiprazole concerning LDL cholesterol (see also Figure 6); from –5.24 (95% CrI: –8.94 to –2.05) for amisulpride to 0.71 (95% CrI: –0.76 to 1.98) for aripiprazole concerning HDL cholesterol (see also Figure 7); and from 38.98 (95% CrI: 12.66 to 66.49) for amisulpride to –11.85 (95% CrI: –28.44 to 4.95) for ziprasidone concerning triglycerides (see also Figure 8).

Table 1.

Map of antipsychotics ranked according to associated alteration in weight gain and metabolic parameters

Weight gain Fasting glucose Total cholesterol LDL cholesterol HDL cholesterol Triglycerides
Fluspirilene LAI –9.13 (–19.02 to 1.43)
Haloperidol LAI –2.53 (–5.32 to 0.26) 0.84 (–11.92 to 13.36) 7.68 (–3.09 to 18.90) 4.00 (–4.67 to 13.15) 0.49 (–2.08 to 3.04) 9.22 (–19.68 to 38.70)
Fluphenazine LAI –1.94 (–5.24 to 1.37)
Fluphenazine –1.30 (–4.45 to 1.93) 15.45 (–19.43 to 51.29)
Ziprasidone –0.16 (–1.15 to 0.84) –0.67 (–5.40 to 4.24) –4.69 (–10.39 to 1.23) –1.32 (–7.32 to 4.38) –0.14 (–1.74 to 1.38) –11.85 (–28.44 to 4.95)
Haloperidol –0.01 (–0.81 to 0.80) 2.72 (–2.32 to 7.96) 2.77 (–3.21 to 8.69) 1.49 (–14.95 to 20.28) –1.58 (–9.94 to 5.91) 6.90 (–13.58 to 27.06)
Placebo 0 0 0 0 0 0
Lurasidone –0.06 (–1.40 to 1.31) 0.96 (–5.33 to 7.43) 3.88 (–3.11 to 11.07) 5.08 (–0.94 to 10.65) 0.70 (–0.97 to 2.54) –13.09 (–33.06 to 7.51)
Aripiprazole LAI –0.00 (–1.08 to 1.08) 2.35 (–1.51 to 6.53) 2.51 (–3.35 to 8.05) 0.60 (–4.06 to 5.49) 0.32 (–1.26 to 1.81) –0.14 (–13.47 to 14.37)
Flupentixol 0.10 (–3.08 to 3.35)
Aripiprazole 0.41 (–0.40 to 1.28) 0.35 (–2.40 to 3.28) –0.75 (–4.90 to 3.21) –1.92 (–5.64 to 1.96) 0.71 (–0.76 to 1.98) –1.07 (–12.26 to 9.87)
Perphenazine 0.61 (–0.76 to 2.01) 4.46 (–3.72 to 12.73) –0.18 (–1.98 to 1.66) 8.79 (–20.83 to 39.49)
Cariprazine 0.62 (–0.82 to 2.05) 1.76 (–2.82 to 6.42) –0.55 (–8.18 to 7.52) 0.73 (–5.51 to 6.97) –1.22 (–3.25 to 0.73) –1.08 (–20.58 to 18.71)
Asenapine 0.73 (–0.32 to 1.81) 3.37 (–0.80 to 7.36) 4.86 (–1.25 to 11.32) 3.25 (–2.93 to 9.67) –0.12 (–1.90 to 1.77) 4.22 (–14.87 to 22.67)
Iloperidone 0.78 (–0.56 to 2.15) –0.24 (–4.40 to 4.57) –0.59 (–9.37 to 8.02) 2.36 (–3.70 to 7.67) –0.33 (–1.80 to 1.27) 14.66 (–3.01 to 29.36)
Amisulpride 1.43 (0.45 to 2.41) 2.13 (–2.72 to 7.04) 9.77 (–6.96 to 26.68) 9.72 (–6.90 to 26.88) –5.24 (–8.94 to –2.05) 38.98 (12.66 to 66.49)
Paliperidone LAI 1.43 (0.55 to 2.33) 0.83 (–2.49 to 4.00) 3.31 (–1.18 to 8.13) 2.29 (–1.62 to 6.35) –0.30 (–1.48 to 0.93) –0.09 (–12.14 to 11.33)
Quetiapine 1.59 (0.79 to 2.42) 3.14 (0.09 to 6.33) 8.20 (3.33 to 13.30) 5.87 (1.33 to 10.51) –1.59 (–2.91 to –0.27) 21.87 (7.79 to 35.81)
Paliperidone 1.73 (0.70 to 2.78) 1.85 (–1.89 to 5.64) 7.58 (2.21 to 13.17) 3.35 (–1.44 to 8.56) 0.15 (–1.42 to 1.75) 4.61 (–8.80 to 18.29)
Brexpiprazole 1.91 (–0.13 to 3.94) 3.62 (–4.37 to 11.71) –0.28 (–14.06 to 13.51) 2.18 (–9.70 to 14.08) –1.31 (–4.31 to 1.70) 2.18 (–24.34 to 28.63)
Risperidone 1.87 (1.12 to 2.65) 3.51 (0.21 to 6.80) 3.62 (–0.93 to 8.28) 4.02 (–0.91 to 9.04) –1.20 (–2.45 to 0.15) 2.88 (–10.54 to 16.07)
Risperidone LAI 2.00 (0.85 to 3.16) 3.34 (–0.38 to 7.21) 7.58 (2.33 to 12.90) 5.84 (0.49 to 11.38) –0.17 (–1.61 to 1.40) 8.40 (–6.63 to 23.83)
Sertindole 2.30 (0.43 to 4.31) 6.44 (–0.21 to 13.06) 9.07 (–6.01 to 24.54) 6.91 (–5.68 to 19.49) 0.24 (–3.42 to 4.61) 8.79 (–18.02 to 35.51)
Olanzapine LAI 3.60 (2.12 to 5.12) 7.64 (3.17 to 13.20) 12.02 (5.07 to 19.01) 9.59 (3.61 to 15.49) –2.91 (–4.45 to –1.18) 20.46 (–0.40 to 41.68)
Pimozide 6.16 (–1.78 to 13.74)
Zotepine 3.87 (2.14 to 5.58)
Olanzapine 3.82 (3.15 to 4.50) 5.07 (2.44 to 7.98) 12.65 (8.73 to 16.51) 8.09 (4.32 to 11.89) –2.59 (–3.71 to –1.44) 31.66 (20.32 to 42.84)
Clozapine 4.21 (3.03 to 5.42) 1.64 (–7.08 to 10.26) 15.83 (–2.44 to 32.73)
Chlorpromazine 5.13 (1.98 to 8.30) 4.94 (–7.93 to 18.90) 13.00 (–2.21 to 29.08)
Open in a new tab

SUCRA value

Numbers present the mean differences (MDs) with their 95% credible intervals (CrIs) from the network meta‐analysis compared to placebo. The order of treatments is according to surface under the cumulative ranking curve (SUCRA) value of the primary outcome “weight gain”. The color gradient from grey to white represents the SUCRA value, with darker fields indicating a higher probability of being the worst drug. Empty cells indicate that no data are available. LAI – long‐acting injectable, LDL – low density lipoprotein, HDL – high density lipoprotein.

Figure 4.

Figure 4

Open in a new tab

Forest plot of antipsychotic drugs vs. placebo for the secondary outcome “fasting glucose”. Network meta‐analysis estimates of treatment effect of each drug vs. placebo are reported as mean differences (MDs) and 95% credible intervals (CrIs). The order of treatments is according to surface under the cumulative ranking curve (SUCRA) ranking. LAI – long‐acting injectable, AMI – amisulpride, ARI – aripiprazole, ASE – asenapine, BRE – brexpiprazole, CAR – cariprazine, CLO – clozapine, CPZ – chlorpromazine, HAL – haloperidol, ILO – iloperidone, LUR – lurasidone, OLA – olanzapine, PAL – paliperidone, PLB – placebo, QUE – quetiapine, RIS – risperidone, SER – sertindole, ZIP – ziprasidone.

Figure 5.

Figure 5

Open in a new tab

Forest plot of antipsychotic drugs vs. placebo for the secondary outcome “total cholesterol”. Network meta‐analysis estimates of treatment effect of each drug vs. placebo are reported as mean differences (MDs) and 95% credible intervals (CrIs). The order of treatments is according to surface under the cumulative ranking curve (SUCRA) ranking. LAI – long‐acting injectable, AMI – amisulpride, ARI – aripiprazole, ASE – asenapine, BRE – brex­piprazole, CAR – cariprazine, CLO – clozapine, CPZ – chlorpromazine, FLU – fluphenazine, HAL – haloperidol, ILO – iloperidone, LUR – lurasidone, OLA – olanzapine, PAL – paliperidone, PER – perphenazine, PLB – placebo, QUE – quetiapine, RIS – risperidone, SER – sertindole, ZIP – ziprasidone.

Figure 6.

Figure 6

Open in a new tab

Forest plot of antipsychotic drugs vs. placebo for the secondary outcome “low‐density lipoprotein (LDL) cholesterol”. Network meta‐analysis estimates of treatment effect of each drug vs. placebo are reported as mean differences (MDs) and 95% credible intervals (CrIs). The order of treatments is according to surface under the cumulative ranking curve (SUCRA) ranking. LAI – long‐acting injectable, AMI – amisulpride, ARI – aripiprazole, ASE – asenapine, BRE – brexpiprazole, CAR – cariprazine, HAL – haloperidol, ILO – iloperidone, LUR – lurasidone, OLA – olanzapine, PAL – paliperidone, PLB – placebo, QUE – quetiapine, RIS – risperidone, SER – sertindole, ZIP – ziprasidone.

Figure 7.

Figure 7

Open in a new tab

Forest plot of antipsychotic drugs vs. placebo for the secondary outcome “high‐density lipoprotein (HDL) cholesterol”. Network meta‐analysis estimates of treatment effect of each drug vs. placebo are reported as mean differences (MDs) and 95% credible intervals (CrIs). The order of treatments is according to surface under the cumulative ranking curve (SUCRA) ranking. LAI – long‐acting injectable, AMI – amisulpride, ARI – aripiprazole, ASE – asenapine, BRE – brexpiprazole, CAR – cariprazine, HAL – haloperidol, ILO – iloperidone, LUR – lurasidone, OLA – olanzapine, PAL – paliperidone, PER – perphenazine, PLB – placebo, QUE – quetiapine, RIS – risperidone, SER – sertindole, ZIP – ziprasidone.

Figure 8.

Figure 8

Open in a new tab

Forest plot of antipsychotic drugs vs. placebo for the secondary outcome “triglycerides”. Network meta‐analysis estimates of treatment effect of each drug vs. placebo are reported as mean differences (MDs) and 95% credible intervals (CrIs). The order of treatments is according to surface under the cumulative ranking curve (SUCRA) ranking. LAI – long‐acting injectable, AMI – amisulpride, ARI – aripiprazole, ASE – asenapine, BRE – brexpiprazole, CAR – cariprazine, HAL – haloperidol, ILO – iloperidone, LUR – lurasidone, OLA – olanzapine, PAL – paliperidone, PER – perphenazine, PLB – placebo, QUE – quetiapine, RIS – risperidone, SER – sertindole, ZIP – ziprasidone.

No evidence of inconsistency was detected for total cholesterol, LDL and HDL cholesterol; little evidence of inconsistency was present for “number of participants with weight gain”, fasting glucose and triglycerides. Heterogeneity for the secondary outcomes ranged between low and low to moderate (see supplementary information).

DISCUSSION

We, for the first time, synthesized the mid‐ to long‐term (median: 45 weeks) evidence on metabolic side effects of 31 antipsychotics in people with schizophrenia, using a network meta‐analysis based on 137 RCTs including 35,007 participants. As antipsychotic drugs are often taken for long periods of time, our results represent more valuable clinical information on these health consequences than previous analyses based on short‐term studies which on average only lasted 6 weeks1, 16.

Every kilogram increase in body weight (our primary outcome) increases the risk of cardiovascular disease by 3.1%16, 46. We found that antipsychotics differ in their propensity to cause weight gain (see Figure 3). For some antipsychotics, the average weight gain was comparable with placebo, in the sense that there was a tendency to either weight loss (fluspirilene LAI, haloperidol LAI and oral, fluphenazine LAI and oral, and ziprasidone) or an average weight gain of up to 1 kg (lurasidone, aripiprazole LAI and oral, flupentixol, perphenazine, cariprazine, asenapine and iloperidone). All CrIs of these drugs included zero, indicating that some weight loss or weight gain is possible.

An average weight gain between 1 and 2 kg compared to pla­cebo was observed for amisulpride, paliperidone LAI and oral, quetiapine, brexpiprazole, and risperidone oral and LAI. An average weight gain higher than 2 kg compared to placebo was estimated for sertindole, olanzapine LAI and oral, pimozide, zotepine, clozapine and chlorpromazine. These drugs with substantial weight gain were also associated with more glucose and lipid disturbances, with olanzapine showing the most pronounced alterations. Of note, for some drugs, the estimates are very uncertain due to small sample sizes, particularly for fluspirilene LAI and pimozide.

In network meta‐regressions, we found no moderating effect for baseline weight, gender, age and ethnicity. Sponsored study arms showed more weight gain compared to non‐sponsored ones, which gives no indication for bias by drug company sponsorship because the effect is not in favour of sponsored drugs.

We found no substantial difference between oral and LAI formulations, and the hierarchy in the sensitivity analysis pooling oral and LAI formulations was similar to the main analysis (see Figure 9). Haloperidol is an exception, since weight loss was observed with its LAI formulation, while the oral formulation was weight neutral. However, haloperidol LAI is only connected to the main network by one study with extreme results 47 , meaning that for this drug, as well as for fluphenazine LAI and fluspirilene LAI, control by indirect evidence is lacking.

Figure 9.

Figure 9

Open in a new tab

Forest plot of antipsychotic drugs vs. placebo for the post‐hoc sensitivity analysis on weight gain pooling long‐acting injectable (LAI) and oral formulations. MD – mean difference, 95% CrI – 95% credible interval, LAI – long‐acting injectable, AMI – amisulpride, ARI – aripiprazole, ASE – asenapine, BRE – brexpiprazole, CAR – cariprazine, CLO – clozapine, CPZ – chlorpromazine, FLP – fluspirilene, FLU – fluphenazine, FPX – flupentixol, HAL – haloperidol, ILO – iloperidone, LUR – lurasidone, OLA – olanzapine, PAL – paliperidone, PER – perphenazine, PIM – pimozide, PLB – placebo, QUE – quetiapine, RIS – risperidone, SER – sertindole, ZIP – ziprasidone, ZOT – zotepine.

The ranking of antipsychotics in all outcomes was comparable with short‐term findings 16 (median treatment duration: 6 weeks vs. 45 weeks here). For fasting glucose and lipid parameters, the magnitude of the effect was also similar. This suggests that the effects on these parameters occur rapidly, and then remain stable.

Weight gain was more pronounced in our mid‐ to long‐term data compared to the reported short‐term data 16 , but not as much as expected, with the highest difference (approximately +1 kg) seen for olanzapine. However, this result is in line with those of other studies: in a pairwise meta‐analysis 26 , a significant additional weight increase after 6 weeks was found only for olanzapine and first‐generation antipsychotics as a group. In a population‐based cohort study in UK primary care 48 , more weight gain was observed during the first 6 weeks of treatment than in the following 4 years. For example, men treated with olanzapine (>5 mg/day) gained on average 4.5 kg in the first six weeks, but only 1.4 kg thereafter. In 573 patients treated with olanzapine for a median of 2.5 years 49 , an average weight gain of 6.26 kg was observed, which plateaued at 39 weeks (compared to 3.82 kg in our meta‐analysis with 6,156 study participants treated with olanzapine for a median of 26 weeks). Taken together, these results suggest that antipsychotic‐induced weight gain stagnates over time5, 48, 49, 50, 51.

Several considerations and limitations need to be taken into account when interpreting our results. First, there is evidence that antipsychotic‐naïve individuals are more vulnerable to weight gain26, 52, but only 11% of our studies included participants with minimal prior exposure to antipsychotics, although excluding these subjects in a sensitivity analysis did not materially change the results.

Second, 140 (older) studies on FGAs and 99 studies on at least one SGA met our inclusion criteria, but did not report weight gain. This missing information led to downgrading the certainty in results with CINeMA, regardless the primary study aim and publication year, although without the original protocol we cannot state whether these outcomes were not measured or not reported.

Third, enriched designs in which patients are stabilized on the drug under investigation before randomization may lead to ceiling effects. Excluding these studies (22/110, 20%) in a sensitivity analysis led to 0.63 kg more weight gain on average, with the most extreme result for iloperidone (1.97 kg versus 0.78 kg in the primary analysis).

Finally, the high dropout rates in long‐term studies are a major concern (42% here). The classical last‐observation‐carried‐forward (LOCF) method underestimates the total weight gain, because the last measurement before dropout is used, which reflects an exposure period shorter than the planned study duration. More sophisticated models such as mixed models of repeated measures (MMRM) try to implement missing data (used by 6/110 studies included here). In our sensitivity analysis including only observed cases, antipsychotics with substantial weight gain in the primary analysis had a somewhat more pronounced effect. Nevertheless, this analysis cannot account for patients who dropped out due to weight gain.

We conclude that antipsychotics differ clearly in weight gain and metabolic parameters in mid‐ to long‐term treatment. The magnitude of the differences in fasting glucose and lipid parameters was approximately the same as previously reported for short‐term studies, suggesting that these effects occur quickly. Differences in weight gain were more pronounced compared to previously published short‐term data. However, the overall evidence seems to suggest that weight gain is most pronounced at the beginning of treatment and then remains somewhat stable. Long‐term studies with initially antipsychotic‐naïve participants are needed.

Although the results were robust to several potential confounders, there was substantial interindividual variability, which could be explored by individual participant data meta‐analysis, and should be considered in treatment decisions.

ACKNOWLEDGEMENTS

This work was supported by the German Ministry for Education and Research (grant no. FKZ 01KG1904). V. Chiocchia and G. Salanti have been supported by the Swiss National Science Foundation (grant no. 179185). The funding sources had no role in the design and conduct of the study, and the preparation, approval and decision to submit the manuscript. The authors thank F. Shokraneh, who conducted the first literature search, F. Krayer for technical support, and all the authors of the included studies, particularly R. Emsley, Y. Koshikawa, L. San and G.D. Kotzalidis, as well as Janssen/Johnson & Johnson (via YODA Project #2020‐4517), Eli Lilly, Vanda and Gedeon Richter, for providing additional data. The interpretation and reporting of data are solely the responsibility of the authors and do not necessarily represent the views of the data sharing agencies. A. Burschinski and J. Schneider‐Thoma contributed equally to this work. Supplementary information on the study is available at https://ebmpp.org/fileadmin/resources/files/Appendix_Metabolic.pdf.

REFERENCES

  • 1. Huhn M, Nikolakopoulou A, Schneider‐Thoma J et al. Comparative efficacy and tolerability of 32 oral antipsychotics for the acute treatment of adults with multi‐episode schizophrenia: a systematic review and network meta‐analysis. Lancet 2019;394:939‐51. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Leucht S, Tardy M, Komossa K et al. Antipsychotic drugs versus placebo for relapse prevention in schizophrenia: a systematic review and meta‐analysis. Lancet 2012;379:2063‐71. [DOI] [PubMed] [Google Scholar]
  • 3. De Hert M, Sermon J, Geerts P et al. The use of continuous treatment versus placebo or intermittent treatment strategies in stabilized patients with schizophrenia: a systematic review and meta‐analysis of randomized controlled trials with first‐ and second‐generation antipsychotics. CNS Drugs 2015;29:637‐58. [DOI] [PubMed] [Google Scholar]
  • 4. Takeuchi H, Suzuki T, Uchida H et al. Antipsychotic treatment for schizophrenia in the maintenance phase: a systematic review of the guidelines and algorithms. Schizophr Res 2012;134:219‐25. [DOI] [PubMed] [Google Scholar]
  • 5. De Hert M, Detraux J, van Winkel R et al. Metabolic and cardiovascular ad­verse effects associated with antipsychotic drugs. Nat Rev Endocrinol 2011;8:114‐26. [DOI] [PubMed] [Google Scholar]
  • 6. Kaur J. A comprehensive review on metabolic syndrome. Cardiol Res Pract 2014;2014:943162. [DOI] [PMC free article] [PubMed] [Google Scholar] [Retracted]
  • 7. Mottillo S, Filion KB, Genest J et al. The metabolic syndrome and cardio­vascular risk a systematic review and meta‐analysis. J Am Coll Cardiol 2010;56:1113‐32. [DOI] [PubMed] [Google Scholar]
  • 8. Correll CU, Solmi M, Veronese N et al. Prevalence, incidence and mortality from cardiovascular disease in patients with pooled and specific severe mental illness: a large‐scale meta‐analysis of 3,211,768 patients and 113,383,368 controls. World Psychiatry 2017;16:163‐80. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Hjorthøj C, Stürup AE, McGrath JJ et al. Years of potential life lost and life expectancy in schizophrenia: a systematic review and meta‐analysis. Lancet Psychiatry 2017;4:295‐301. [DOI] [PubMed] [Google Scholar]
  • 10. Allison DB, Mackell JA, McDonnell DD. The impact of weight gain on quality of life among persons with schizophrenia. Psychiatr Serv 2003;54:565‐7. [DOI] [PubMed] [Google Scholar]
  • 11. Perkins DO. Predictors of noncompliance in patients with schizophrenia. J Clin Psychiatry 2002;63:1121‐8. [DOI] [PubMed] [Google Scholar]
  • 12. Weiden PJ, Mackell JA, McDonnell DD. Obesity as a risk factor for antipsychotic noncompliance. Schizophr Res 2004;66:51‐7. [DOI] [PubMed] [Google Scholar]
  • 13. Tandon R, Belmaker RH, Gattaz WF et al. World Psychiatric Association Pharmacopsychiatry Section statement on comparative effectiveness of antipsychotics in the treatment of schizophrenia. Schizophr Res 2008;100:20‐38. [DOI] [PubMed] [Google Scholar]
  • 14. National Collaborating Centre for Mental Health . Psychosis and schizophrenia in adults: treatment and management. London: National Institute for Health and Care Excellence, 2014. [PubMed] [Google Scholar]
  • 15. American Psychiatric Association . The American Psychiatric Association practice guideline for the treatment of patients with schizophrenia. Washington: American Psychiatric Association Publishing, 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16. Pillinger T, McCutcheon RA, Vano L et al. Comparative effects of 18 antipsychotics on metabolic function in patients with schizophrenia, predictors of metabolic dysregulation, and association with psychopathology: a systematic review and network meta‐analysis. Lancet Psychiatry 2020;7:64‐77. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Schneider‐Thoma J, Kapfhammer A, Wang D et al. Metabolic side effects of antipsychotic drugs in individuals with schizophrenia during medium‐ to long‐term treatment: protocol for a systematic review and network meta‐analysis of randomized controlled trials. Syst Rev 2021;10:214. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18. Cochrane Schizophrenia . https://schizophrenia.cochrane.org/.
  • 19. Tong Z, Li F, Ogawa Y et al. Quality of randomized controlled trials of new generation antidepressants and antipsychotics identified in the China National Knowledge Infrastructure (CNKI): a literature and telephone interview study. BMC Med Res Methodol 2018;18:96. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20. Woodhead M. 80% of China's clinical trial data are fraudulent, investigation finds. BMJ 2016;355:i5396. [DOI] [PubMed] [Google Scholar]
  • 21. Parry J. China vows to clamp down on academic fraud amid medical journal scandal. BMJ 2017;357:j2970. [DOI] [PubMed] [Google Scholar]
  • 22. Leucht S, Davis JM. Which first‐generation antipsychotics should be “repurposed” for the treatment of schizophrenia. Eur Arch Psychiatry Clin Neurosci 2022;272:1‐3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23. Misawa F, Kishimoto T, Hagi K et al. Safety and tolerability of long‐acting injectable versus oral antipsychotics: a meta‐analysis of randomized controlled studies comparing the same antipsychotics. Schizophr Res 2016;176:220‐30. [DOI] [PubMed] [Google Scholar]
  • 24. Correll CU, Lencz T, Malhotra AK. Antipsychotic drugs and obesity. Trends Mol Med 2011;17:97‐107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25. Gardner DM, Murphy AL, O'Donnell H et al. International consensus study of antipsychotic dosing. Am J Psychiatry 2010;167:686‐93. [DOI] [PubMed] [Google Scholar]
  • 26. Bak M, Fransen A, Janssen J et al. Almost all antipsychotics result in weight gain: a meta‐analysis. PLoS One 2014;9:e94112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27. Zhang J‐P, Gallego JA, Robinson DG et al. Efficacy and safety of individual second‐generation vs. first‐generation antipsychotics in first‐episode psychosis: a sys­tematic review and meta‐analysis. Int J Neuropsychopharmacol 2013;16:1205‐18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28. Spertus J, Horvitz‐Lennon M, Abing H et al. Risk of weight gain for specific antipsychotic drugs: a meta‐analysis. NPJ Schizophr 2018;4:12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Barton BB, Segger F, Fischer K et al. Update on weight‐gain caused by antipsychotics: a systematic review and meta‐analysis. Expert Opin Drug Saf 2020;19:295‐314. [DOI] [PubMed] [Google Scholar]
  • 30. Rummel‐Kluge C, Komossa K, Schwarz S 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;123:225‐33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Pagsberg AK, Tarp S, Glintborg D et al. Acute antipsychotic treatment of children and adolescents with schizophrenia‐spectrum disorders: a systematic review and network meta‐analysis. J Am Acad Child Adolesc Psychiatry 2017;56:191‐202. [DOI] [PubMed] [Google Scholar]
  • 32. De Hert M, Yu W, Detraux J et al. Body weight and metabolic adverse effects of asenapine, iloperidone, lurasidone and paliperidone in the treatment of schizophrenia and bipolar disorder: a systematic review and exploratory meta‐analysis. CNS Drugs 2012;26:733‐59. [DOI] [PubMed] [Google Scholar]
  • 33. Zhang Y, Liu Y, Su Y et al. The metabolic side effects of 12 antipsychotic drugs used for the treatment of schizophrenia on glucose: a network meta‐analysis. BMC Psychiatry 2017;17:373. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Sterne JAC, Savović J, Page MJ et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ 2019;2:l4898. [DOI] [PubMed] [Google Scholar]
  • 35. Turner RM, Jackson D, Wei Y et al. Predictive distributions for between‐study heterogeneity and simple methods for their application in Bayesian meta‐analysis. Stat Med 2015;34:984‐98. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36. Rhodes KM, Turner RM, Higgins JPT. Predictive distributions were developed for the extent of heterogeneity in meta‐analyses of continuous outcome data. J Clin Epidemiol 2015;68:52‐60. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37. Dias S, Welton NJ, Caldwell DM et al. Checking consistency in mixed treatment comparison meta‐analysis. Stat Med 2010;29:932‐44. [DOI] [PubMed] [Google Scholar]
  • 38. Higgins JPT, Jackson D, Barrett JK et al. Consistency and inconsistency in network meta‐analysis: concepts and models for multi‐arm studies. Res Synth Methods 2012;3:98‐110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39. Chaimani A, Salanti G. Using network meta‐analysis to evaluate the existence of small‐study effects in a network of interventions. Res Synth Methods 2012;3:161‐76. [DOI] [PubMed] [Google Scholar]
  • 40. Peters JL, Sutton AJ, Jones DR et al. Contour‐enhanced meta‐analysis funnel plots help distinguish publication bias from other causes of asymmetry. J Clin Epidemiol 2008;61:991‐6. [DOI] [PubMed] [Google Scholar]
  • 41. Béliveau A, Boyne DJ, Slater J et al. BUGSnet: an R package to facilitate the conduct and reporting of Bayesian network meta‐analyses. BMC Med Res Methodol 2019;19:196. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42. Plummer M. JAGS: a program for analysis of Bayesian graphical models using Gibbs sampling. Presented at the 3rd International Workshop on Distributed Statistical Computing, Vienna, March: 2003. [Google Scholar]
  • 43. Rücker G, König J, Efthimiou O et al. Package ‘netmeta’: network meta‐analysis using frequentist methods. https://cran.r‐project.org/web/packages/netmeta/​netmeta.pdf.
  • 44. Schwarzer G. Package ‘meta’: general package for meta‐analysis. https://cran.r‐project.org/web/packages/meta/meta.pdf.
  • 45. Nikolakopoulou A, Higgins JPT, Papakonstantinou T et al. CINeMA: an approach for assessing confidence in the results of a network meta‐analysis. PLoS Med 2020;17:e1003082. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46. Willett WC, Manson JE, Stampfer MJ et al. Weight, weight change, and coronary heart disease in women. Risk within the ‘normal’ weight range. JAMA 1995;273:461‐5. [DOI] [PubMed] [Google Scholar]
  • 47. McEvoy JP, Byerly M, Hamer RM et al. Effectiveness of paliperidone palmitate vs haloperidol decanoate for maintenance treatment of schizophrenia: a randomized clinical trial. JAMA 2014;311:1978‐87. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48. Bazo‐Alvarez JC, Morris TP, Carpenter JR et al. Effects of long‐term antipsychotics treatment on body weight: a population‐based cohort study. J Psychopharmacol 2020;34:79‐85. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49. Kinon BJ, Basson BR, Gilmore JA et al. Long‐term olanzapine treatment: weight change and weight‐related health factors in schizophrenia. J Clin Psychiatry 2001;62:92‐100. [PubMed] [Google Scholar]
  • 50. Millen BA, Campbell GM, Beasley CM. Weight changes over time in adults treated with the oral or depot formulations of olanzapine: a pooled analysis of 86 clinical trials. J Psychopharmacol 2011;25:639‐45. [DOI] [PubMed] [Google Scholar]
  • 51. Pérez‐Iglesias R, Martínez‐García O, Pardo‐Garcia G et al. Course of weight gain and metabolic abnormalities in first treated episode of psychosis: the first year is a critical period for development of cardiovascular risk factors. Int J Neuropsychopharmacol 2014;17:41‐51. [DOI] [PubMed] [Google Scholar]
  • 52. Alvarez‐Jiménez M, González‐Blanch C, Crespo‐Facorro B et al. Antipsychotic‐induced weight gain in chronic and first‐episode psychotic disorders: a systematic critical reappraisal. CNS Drugs 2008;22:547‐62. [DOI] [PubMed] [Google Scholar]

Articles from World Psychiatry are provided here courtesy of The World Psychiatric Association

RESOURCES