Introduction

Schizophrenia is a severe and complex psychiatric disorder, characterized by psychotic symptoms (e.g., hallucinations, delusions, and disorganization) and motivational and cognitive dysfunctions1. The global prevalence of schizophrenia is estimated to increase from 14.2 million in 1990 to 23.6 million in 2019, corresponding to a relative increase of 65.9%2. According to a cross-sectional analysis in China, the 12-month prevalence of schizophrenia was 0.6%3. Schizophrenia is associated with adverse health outcomes, including physical comorbidities4, disability2, and premature mortality5. Metabolic syndrome, a combination of metabolic defects comprising abdominal obesity, insulin resistance, dysglycemia, atherogenic dyslipidemia, and hypertension6, is regarded as a common comorbidity in patients with schizophrenia. A previous systematic review of 77 studies concluded that the overall prevalence of metabolic syndrome was 32.5% (95% CI = 30.1%–35.0%) in schizophrenia, and the rates of some metabolic abnormalities (e.g., overweight, hyperglycemia and lipid abnormalities) were disproportionately high in those diagnosed with schizophrenia7. Comorbid metabolic syndrome in patients with schizophrenia increased the risk of cardiovascular diseases8, cognitive impairment9, and mortality4, presented considerable social, economic, and healthcare challenges.

Antipsychotics, especially second-generation antipsychotics (SGAs), have been widely acknowledged to be effective treatments in the management of schizophrenia10,11. Compared to the first-generation antipsychotics, SGAs present better effectiveness on the negative, cognitive, and mood symptoms of schizophrenia and result in fewer extrapyramidal symptoms10. However, increasing evidence indicated that SGAs elevated the risk of metabolic abnormalities in patients with schizophrenia through neurotransmitters, hormones involved in the management of satiety, feeding, and glucose metabolism12. A systematic review comparing 18 antipsychotics on metabolic function in patients with schizophrenia from 100 randomized controlled trials revealed that olanzapine and clozapine exhibited the highest metabolic side-effects, whereas aripiprazole, brexpiprazole, cariprazine, lurasidone, and ziprasidone had the most benign side-effects13. A cross-sectional analysis of 833 psychiatric patients found that receiving olanzapine is a significant indicator of metabolic syndrome14. An observational study of 60 patients with schizophrenia reported that groups prescribed risperidone and clozapine experienced a higher likelihood of metabolic syndrome than the group prescribed aripiprazole15. Another study using data of 254 patients with schizophrenia found that the proportions of diagnosed hypertension, diabetes, and dyslipidemia were significantly higher in the clozapine group than in the non-clozapine group16. These studies suggested that different SGAs were associated with metabolic abnormalities to varying extents, and the choice of SGAs should consider the clinical circumstances and the preferences of patients, carers, and clinicians13.

In 2020, Chinese Schizophrenia Coordination Group released the expert consensus on the management of metabolic syndrome in patients with schizophrenia and specified the metabolic risk of SGAs: 1) clozapine and olanzapine with a high risk; 2) amisulpride, quetiapine, and risperidone with an intermediate risk; and 3) aripiprazole, ziprasidone, and lurasidone with a low risk17. This expert consensus suggested adding or switching SGAs with a low metabolic risk for patients with schizophrenia and metabolic abnormalities while delivering lifestyle modifications. The mitigating effects of switching antipsychotics on weight gain in patients with severe mental illness have been confirmed in a previous systematic review18. However, no local attempts have been made to examine whether using low-risk SGAs would alleviate metabolic syndrome or metabolic abnormalities in patients with schizophrenia. As antipsychotic medications are essential for treating schizophrenia, to produce evidence on the associations between low-risk SGAs in patients with schizophrenia and metabolic outcomes would help clinicians design appropriate treatment that targets both schizophrenia and metabolic abnormalities, with important implications for disease prognosis, and improving their quality of life and social function. Therefore, this study used provincial representative samples of patients with both schizophrenia and metabolic abnormalities and examined the associations of adding or switching to low-risk SGAs with alleviating metabolic syndrome and metabolic abnormalities.

Methods

Study design and participants

This study was a real-world observational analysis using data of patients with schizophrenia from the Yihua research platform (https://eyan.yihuayidong.com/). Data was collected at 25 healthcare organizations across Zhejiang Province, China. This platform encompasses individual-level data including demographic characteristics, disease diagnoses, prescribed medications, and metabolic parameters. Ethics approvals were received from the Institution Review Board at Zhejiang Provincial People’s Hospital (KY2022027). All informed participants provided written consent.

The metabolic risk of SGAs was identified into three levels: 1) clozapine and olanzapine with a high risk; 2) amisulpride, quetiapine and risperidone with an intermediate risk; and 3) aripiprazole, ziprasidone, and lurasidone with a low risk. Adult patients were included if they 1) ≥18 years; 2) were diagnosed with schizophrenia based on ICD-10 criteria; 3) have not take SGAs with a low metabolic risk yet; and 4) had at least one metabolic abnormality. According to the Chinese Guideline for the Prevention and Treatment of Type 2 Diabetes Mellitus, the metabolic abnormalities were determined via the following criteria: 1) waist circumference ≥90 cm in men and in ≥85 cm in women, 2) fasting blood glucose ≥ 6.1 mmol/L, 3) blood pressure ≥130/85 mmHg, 4) triglyceride ≥1.70 mmol/L, and 5) high-density lipoprotein cholesterol (HDL-C) < 1.04 mmol/L. This guideline is in accordance with other well-established disgnostic criteria [e.g., the National Cholesterol Education Program Adult Treatment Panel III criteria (NCEP ATP III criteria) and the International Diabetes Federation criteria], and is more suitable for Chinese. The exclusion criteria were as follows: having organic brain diseases or serious physical diseases, having psychoactive substance dependence, having mental retardation, uncooperative or at risk (e.g., extreme excitement, stupefaction, and passive suicide), and in pregnancy or lactation. The baseline was defined as the first record of SGAs after preliminary screening between January 1, 2022 and December 31, 2023.

Outcomes

The primary outcomes were the proportion of participants with metabolic syndrome and the number of metabolic abnormalities at 4 weeks, 8 weeks, 3 months and 6 months after baseline. The number of metabolic abnormalities ranged from 0 to 5, and metabolic syndrome was defined as having three or more metabolic abnormalities. The secondary outcomes were the changes of metabolic parameters (weight, waist circumference, systolic blood pressure, diastolic blood pressure, glucose, triglyceride and HDL-C) and the severity of schizophrenia. The severity of schizophrenia was assessed by the Clinical Global Impression-Severity (CGI-S), with a higher score indicating a higher severity of the disease. The CGI-S scale is concise and easy to apply in a clinical setting, based upon a rapid assessment of all dimensions of symptomatology in schizophrenia19. Previous studies have proved the robust correlation between CGI-S and other validated scales in observational studies and clinical trials for schizophrenia19,20.

Statistical analysis

A 1:1 propensity score matching (PSM) using a nearest-neighbor method with a caliper width of 0.3 standard deviation of the logit propensity score was conducted. A propensity score model was developed using logistic regression with baseline characteristics21. The prescription of SGAs with different metabolic risks was specified as the dependent variable, and characteristics at baseline including sociodemographic characteristics (age, sex, marital status, occupation and educational level), lifestyle factors (smoking status and alcohol consumption), history of physical conditions (locomotor, nervous, hormonal, circulatory, respiratory, digestive, urinary and reproductive system diseases), duration of schizophrenia (years), CGI-S score, polypharmacy, and the baseline levels of metabolic parameters were used as independent variables. Continuous variables were presented as mean (± standard deviation), and categorical variables were presented as number (percentage) for both groups before and after PSM. Chi-square tests and Student’s t-tests were used to compare the between-group differences before PSM, and paired t-tests were used for continuous variables after PSM.

The within-group and between-group risk difference (RDs) and 95% confidence intervals (CIs) in the proportion of participants with metabolic syndrome, and the mean differences (MDs) and 95% CIs in the number of metabolic abnormalities were calculated at 4 weeks, 8 weeks, 3 months, and 6 months. Generalized estimating equations (GEEs) were used to evaluate changes in proportion of participants with metabolic syndrome (binomial distribution and logit link) and the number of metabolic abnormalities (Poisson distribution and log link) associated with the low-risk SGA group compared with the high/intermediate-risk SGAs group. The dose of SGAs was adjusted in the models.

For secondary outcomes, the within-group and between-group differences were compared using paired t-tests, with the mean differences and 95% CIs between the low-risk SGA group and the high/intermediate-risk SGAs group calculated. Linear mixed models with random effects of intercept and time to determine the association between prescribing SGAs and the level of metabolic parameters and CGI-S score at 4 weeks, 8 weeks, 3 months, and 6 months, adjusting for the dose of SGAs. Participants prescribed only one SGA before PSM were further categorized to compare association variations in different SGAs. The group prescribing lurasidone was excluded due to the limited sample size (n = 2).

To test the robustness of the association between low-risk SGAs and metabolic outcomes, we conducted a sensitivity analysis by define metabolic syndrome and metabolic abnormalities using the NCEP ATP III criteria (for details of the NCEP ATP III Criteria, see Table S1). Subgroup analyses by age, sex, physical conditions, body mass index (BMI), duration, lifestyles (smoking or drinking) and polypharmacy were performed to examine the potential variations in the associations between low-risk SGAs and metabolic outcomes. The cut-off values of age, BMI, and duration of schizophrenia were based on the average values of include participants. We additionally separated the low-risk SGAs group into the “adding” (n = 128) and “switching” group (n = 95) and compared the associations between different interventions and metabolic outcomes.

Two-sided p-value of <0.05 was used for all analysis for the statistical significance. Statistical analyses were performed with SAS (Version 9.4) and R (Version 4.5.0).

Results

Participants selection and characteristics

Among the 2189 participants diagnosed with schizophrenia were recruited in the preliminary screening, 658 (30.1%) with no metabolic abnormalities, 264 (12.1%) already prescribed with low-risk SGAs, and 503 (23.0%) with missing data on metabolic syndrome components were excluded (Fig. 1). A total of 764 participants were included, with 279 (36.5%) prescribed SGAs with a low metabolic risk, and 485 (63.5%) prescribed SGAs with a high/intermediate metabolic risk met the study criteria and were included in the analysis. The mean [standard deviation (SD)] age was 50.0 (12.4) years, 288 (37.7%) were women, and the mean duration of schizophrenia was 29.7 (11.8) years. Participants prescribed low-risk SGAs were more likely to be females, have no occupation, higher body mass index (BMI) and metabolic abnormalities, and be prescribed multiple SGAs (Table 1). Age, marital status, educational level, history of physical conditions, lifestyle factors, and the duration and CGI-S score were similar between groups (p > 0.05).

Fig. 1: Flow of study population.
Fig. 1: Flow of study population.The alternative text for this image may have been generated using AI.
Full size image

ICD-10, International Classification of Diseases 10th Revision; SGA, second-generation antipsychotic.

Table 1 Baseline characteristics of participants with schizophrenia before and after propensity score matching.
Full size table

Primary outcomes

The 1:1 propensity-matched algorithm including participants who prescribed low-risk and high/intermediate SGAs yielded 223 matched pairs and accounted for 79.9% (223 of 279) of the participants with SGAs. P-values for all baseline characteristics were >0.05.

In the propensity-matched cohort, the proportion of participants with metabolic syndrome decreased from 8 weeks after baseline to 6 months (8 weeks vs. baseline: RD = −15.70%, 95% CI = −24.71% to −6.68%; 3 months vs. baseline: RD = −15.25%, 95% CI = −24.27% to −6.22%; 6 months vs. baseline: RD = −18.83%, 95% CI = −27.74% to −9.93%, Fig. 2A) in the low-risk SGAs group, whereas the changes in the proportion of participants with metabolic syndrome of high/intermediate risk SGAs group did not reach significance (Fig. 2B). The proportion of participants with metabolic syndrome was significantly lower in the low-risk SGAs group than in the high/intermediate-risk SGAs group at 8 weeks (RD = −16.14%, 95% CI = −25.16% to −7.13%), 3 months (RD = −11.21%, 95% CI = −20.21% to −2.21%), and 6 months (RD = −19.28%, 95% CI = −28.19% to −10.38%, Fig. 2C). The generalized estimated equations also showed differences between the low-risk and high/intermediate-risk SGAs groups, with the odds ratios for metabolic syndrome of 0.54 (95% CI = 0.37–0.78), 0.66 (95% CI = 0.43–1.00) and 0.47 (95% CI = 0.30–0.72) at 8 weeks, 3 months, and 6 months after baseline (Table 2).

Fig. 2: Changes in primary outcomes from baseline to 4 weeks, 8 weeks, 3 months and 6 months in the matched population.
Fig. 2: Changes in primary outcomes from baseline to 4 weeks, 8 weeks, 3 months and 6 months in the matched population.The alternative text for this image may have been generated using AI.
Full size image

A Within-group changes in the proportion of patients with metabolic syndromes in the group of low-risk SGAs. B Within-group changes in the proportion of patients with metabolic syndromes in the group of high/intermediate-risk SGAs. C Between-group changes in the proportion of patients with metabolic syndromes. Changes are estimated as risk differences with 95% confidence intervals by chi square tests. D Within-group changes in the number of metabolic abnormalities in the group of low-risk SGAs. E Within-group changes of the number of metabolic abnormalities in the group of high/intermediate-risk SGAs. F Between-group changes of the number of metabolic abnormalities. Changes are estimated as mean differences with 95% confidence intervals by paired t-tests. SGA second-generation antipsychotics, RD risk difference, MD mean difference.

Table 2 Linear mixed models of metabolic components and CGI-S score by groups in the matched population.
Full size table

For the number of metabolic abnormalities, the low-risk SGAs group showed an consistent decrease from baseline (4 weeks vs. baseline: MD = –0.21, 95% CI = −0.32 to −0.10; 8 weeks vs. baseline: MD = −0.42, 95% CI = −0.54 to −0.30; 3 months vs. baseline: MD = −0.49, 95% CI = −0.63 to −0.35; 6 months vs. baseline: MD = −0.57, 95% CI = −0.72 to −0.43, Fig. 2D), whereas the high/intermediate-risk SGAs group did not show significant differences in most time points compared to baseline (Fig. 2E). The number of metabolic abnormalities was significantly lower in the low-risk SGAs group than in the high/intermediate-risk SGAs group at 8 weeks (MD = −0.33, 95% CI = −0.52 to −0.13), 3 months (MD = −0.37, 95% CI = −0.58 to −0.16), and 6 months (MD = −0.52, 95% CI = −0.75 to −0.29, Fig. 2F). The generalized estimated equations also showed differences between the low-risk and high/intermediate-risk SGAs groups, with the relative number of metabolic abnormalities 14% (relative risk (RR) = 0.86, 95% CI = 0.80 to 0.93), 15% (RR = 0.85, 95% CI = 0.77 to 0.93), and 21% (RR = 0.79, 95% CI = 0.71 to 0.87) lower in the low-risk SGAs group compared to the high/intermediate-risk SGAs group at 8 weeks, 3 months and 6 months, respectively (Table 2).

The proportion of participants with metabolic syndrome and number of metabolic abnormalities at 4 weeks, 8 weeks, 3 months and 6 months according to the type of SGAs were displayed in Supplementary Figures S1, S2. There were 122 (28.8%), 95 (22.5%), 59 (13.9%), 37 (8.7%), 13 (3.1%), 10 (2.3%), and 87 (20.6%) participants were prescribed olanzapine, clozapine, quetiapine, risperidone, amisulpride, aripiprazole, and ziprasidone, respectively. Participants with ziprasidone had a consistently lower proportion of metabolic syndrome and less metabolic abnormalities over time (p < 0.05).

Sensitivity analysis using the NCEP ATP III criteria showed similar patterns of the associations between switching or adding low-risk SGAs and metabolic syndrome and metabolic abnormalities (Supplementary Fig. S3 and Table S2). Subgroup analyses suggested that the associations between low-risk SGAs and metabolic outcomes were more pronounced in patients with no comorbid physical conditions, lower BMI, smoking or drinking and shorter duration (Supplementary Tables S3 & S4). After separating the low-risk SGAs group into the “adding” (n = 128) and “switching” group (n = 95), we found that participants in the switching group were more likely to be females and had higher BMI, blood pressure and glucose (Supplementary Tables S5). Both switching and adding interventions were associated with decreased proportion of metabolic syndrome and number of metabolic abnormalities (Supplementary Fig. S4), while the “adding” group showed significant improvement in more metabolic parameters (Supplementary Table S6).

Secondary outcomes

The changes in metabolic parameters and CGI-S score from baseline to 6 months of two groups were shown in Supplementary Table S7. In the low-risk SGAs group, participants had a higher level of HDL-C at 4 weeks (1.2 ± 0.6 mmol/L), 8 weeks (1.2 ± 0.5 mmol/L), 3 months (1.2 ± 0.6 mmol/L), and 6 months (1.2 ± 0.6 mmol/L), compared to that at baseline. From 8 weeks after baseline, the levels of weight and triglyceride were consistently lower than those at baseline (weight: 69.1 ± 12.1 kg, triglyceride: 1.8 ± 0.9 mmol/L at 8 weeks; weight: 69.0 ± 12.0 kg, triglyceride: 1.7 ± 0.8 mmol/L at 3 months; weight: 68.4 ± 11.6 kg, triglyceride: 1.7 ± 0.8 mmol/L at 6 months). The level of glucose significantly decreased only at 3 months compared to that at baseline (5.3 ± 1.0 mmol/L). In the high/intermediate SGAs group, no significant improvement of metabolic parameters was observed from baseline to 6 months. In addition, systolic blood pressure significantly increased over time in both groups. The level of triglyceride were significantly lower in the low-risk SGAs group than in the high/intermediate risk SGAs group at 8 weeks (MD = –0.25 mmol/L, 95% CI = –0.45 to –0.06 mmol/L), 3 months (MD = –0.35 mmol/L, 95% CI = –0.55 to –0.15 mmol/L) and 6 months (MD = –0.41 mmol/L, 95% CI = –0.61 to –0.21 mmol/L). The level of weight was significantly lower in the low-risk SGAs group than in the high/intermediate risk SGAs group at 6 months (MD = 3.31 kg, 95% CI = −5.48 to −1.13 kg). Both groups showed significant improvement in the CGI-S score from baseline to all time points (the high/intermediate-risk SGAs group: 3.9 ± 1.0; the low-risk SGAs group: 3.7 ± 1.1 at 6 months). From 4 weeks to 3 months after baseline, participants in the low-risk SGAs group had a significantly lower CGI-S score than those in the high/intermediate-risk SGAs group (MD = −0.34, 95% CI = −0.51 to −0.17 at 4 weeks; MD = −0.35, 95% CI = −0.54 to −0.17 at 8 weeks; MD = −0.27, 95% CI = −0.47 to −0.06).

In Table 2, results from linear mixed models showed that SGAs with a low metabolic risk were associated with lower levels of weight (β = −1.525, 95% CI = −2.433 to −0.617 at 8 weeks; β = −1.937, 95% CI = −2.845 to −1.028 at 3 months; β = −3.153, 95% CI = −4.061 to −2.245 at 6 months), glucose (β = −0.208, 95% CI = −0.382 to −0.033 at 6 months) and triglyceride (β = −0.215, 95% CI = −0.360 to −0.071 at 8 weeks; β = −0.313, 95% CI = −0.458 to −0.169 at 3 months; β = &−0.374, 95% CI = −0.518 to −0.230 at 6 months). The CGI-S score was also significantly lower in the low-risk SGAs group than in the high/intermediate SGAs group from 4 weeks to 3 months after baseline (β = −0.234, 95% CI = −0.445 to −0.024 at 3 months, Fig. 3).

Fig. 3: Mean predicted values (95% confidence intervals) of metabolic parameters and CGI-S score by groups in the matched population.
Fig. 3: Mean predicted values (95% confidence intervals) of metabolic parameters and CGI-S score by groups in the matched population.The alternative text for this image may have been generated using AI.
Full size image

Adjusted for dose of SGAs. SBP systolic blood pressure, DBP diastolic blood pressure, HDL-C high density lipoprotein cholesterol, CGI-S Clinical Global Impressions scale-Severity, SGA second-generation antipsychotics.

Discussion

In this real-world observational study, we included participants with schizophrenia and metabolic abnormalities and examined the effectiveness and safety of adding or switching low-risk SGAs. During a 6-month follow-up, the proportion of participants with metabolic syndrome and the absolute number of metabolic abnormalities in the low-risk SGAs group consistently decreased, showing significant differences compared to the high/intermediate SGAs group between 8 weeks and 6 months after baseline. The low-risk SGAs group presented significant improvement in metabolic parameters, including lower levels of weight, glucose and triglyceride between 8 weeks and 6 months after baseline, compared to the high/intermediate risk SGAs group. Both groups exhibited a decreased CGI-S score over time, and the decrease in the low-risk SGAs group was significantly stronger than in the high/intermediate SGAs group between 4 weeks and 3 months after baseline.

Previous studies have reported on the association between SGAs and metabolic abnormalities. For example, a randomized, multi-center, pharmacologic trial conducted in 32 hospitals across China found that antipsychotics influenced changes in BMI, waist circumference, systolic blood pressure, glucose, and triglyceride22. The metabolic side-effects of SGAs might be explained by abnormal glucolipid metabolism from several receptor pathways (e.g., histamine H1, serotonin, D2 dopamine), resulting in increased appetite, weight gaining and metabolic dysfunction23,24. A growing body of literature has further suggested that different SGAs had varying metabolic side-effects for patients with schizophrenia13,14,15,16. A cross-sectional analysis of 833 patients with schizophrenia, schizoaffective and bipolar disorders revealed diastolic blood pressure and receiving olanzapine as significant predictors for developing metabolic syndrome14. Based on previous research and treatment guidelines25,26, the Chinese expert consensus on the management of metabolic syndrome in patients with schizophrenia in 2020 specified the metabolic risk of a range of SGAs, including clozapine and olanzapine with a high risk; quetiapine and risperidone with an intermediate risk; and aripiprazole, ziprasidone, and lurasidone with a low risk. Furthermore, this expert consensus provided important secondary prevention strategies for patients with both schizophrenia and metabolic abnormalities, containing lifestyles interventions and adding or switching to low-risk SGAs. In many cases, patients with schizophrenia and metabolic abnormalities exhibited low adherence to the lifestyle modifications27,28,29, therefore adding or switching to low-risk SGAs might be a better strategy for them. However, no studies have been conducted to examine whether and to what extent adding or switching to low-risk SGAs would mitigate metabolic abnormalities for patients with schizophrenia. In addition, switching antipsychotics for patients with schizophrenia might result in illness exacerbation and new adverse effects, suggesting making a decision to switch antipsychotics should consider the clinical environment, dosing regimen and patient or carer preferences30,31,32. Here we conducted a real-world observational study in 764 patients with both schizophrenia and metabolic abnormalities and dichotomized them into the low-risk SGAs group and the high/intermediate SGAs group, according to whether adding or switching to low-risk SGAs in the clinical environment. After 1:1 PSM, 223 matched pairs were yielded, with non-significant differences in baseline characteristics between groups. We observed a significant decrease in the proportion of patients with metabolic syndrome and the absolute number of metabolic abnormalities in the low-risk SGAs group, compared to the high/intermediate-risk SGAs group. Results from generalized estimated equations showed similar patterns from baseline to 6 months, confirming the strategy of adding or switching to low-risk SGAs could mitigate metabolic syndrome and metabolic abnormalities for patients with schizophrenia. Subgroup analysis suggested that patients with no physical conditions, lower BMI, smoking and drinking and shorter duration can be benefited more in the intervention of switching or adding SGAs. In addition, we found low-risk SGAs were related to the decrease in weight, glucose, and triglyceride. Previous studies only examine the metabolic side-effects of antipsychotics in patients with schizophrenia, regardless of their metabolic parameters at recruitment, limiting comparison with our studies. It is proposed that SGAs with low metabolic risk were associated with a lower blockade at H1 receptors, which was directly proportional to weight gain33. In addition, SGAs with low metabolic risk have inhibition effects on synaptic reuptake of serotonin and norepinephrine, resulting in slight effect on weight34. SGAs with high/intermediate metabolic risk, e.g., olanzapine and risperidone, can impair lysosomal function affecting autophagy and autophagosome clearance or increase lipid accumulation, while ziprasidone, a low-risk SGA, can increase the autophagic flux and reduce intracellular lipids through activating adenosine monophosphate-dependent kinase35. Comparing studies from other countries, the weight and BMI of participants in our analysis was slightly lower36,37,38. However, the BMI of participants in our analysis was comparable with another study of Chinese population, suggesting our study covered a representative sample of patient with schizophrenia and metabolic abnormalities in China39. The differences in weight and BMI between Southern Chinese population and Wastern country population might be explained by the diverse dietary patterns. The Western dietary pattern, characterised by high intakes of red meats, eggs, seafood, cheese, fast foods, snacks, chocolates, alcoholic beverages and coffee, has been widely observed to be associated with metabolic syndrome and disorders40,41,42. While in Zhejiang Province in China, the dietary pattern is characterised by high intakes of refined grains, vegetables, fruits, pickles vegetables, fish and shrimo, bacon and salted fish, salted and preserved eggs, milk, soya bean and its products, miscellaneous bean, fats, drinks40,43. Considering the association between metabolic abnormalities in patients with schizophrenia and the risk of cardiovascular and other morbidity, our findings have important clinical implications for tailored treatment for patients already with metabolic abnormalities, thereby improving their disease diagnosis, quality of life and social function. After separating the low-risk SGAs group into the “adding” and “switching” group, we further observed the adding group showed improvement in more metabolic parameters. This indicated that adding low-risk SGAs could be more effective when switching high/intermediate-risk SGAs showed no improvement.

Eight SGAs, including olanzapine, clozapine, quetiapine, risperidone, amisulpride, aripiprazole, ziprasidone, and lurasidone, were considered in our analysis. Of these, aripiprazole, ziprasidone, and lurasidone were of a low metabolic risk, and the remaining a high/intermediate risk. We further used the subsample prescribed only one SGA to specify the association of each SGA with the change in metabolic syndrome and metabolic abnormalities. We observed non-significant fluctuations in the proportion of patients with metabolic syndrome in most groups, while the ziprasidone group exhibited a consistent decrease from baseline to 6 months. Previous studies have suggested ziprasidone was one of the antipsychotics with the most benign metabolic side-effects13. Zhang et al assigned 2550 patients with schizophrenia into 7 antipsychotic groups and found that the risk of metabolic syndrome is highest in the group prescribed olanzapine, followed by quetiapine, perphenazine, risperidone, and is lowest in the aripiprazole, haloperidol and ziprasidone groups22. Another multi-country study compared the effectiveness and costs of ten antipsychotics and found that the lurasidone and ziprasidone groups had the lowest incidence of diabetes and cardiovascular diseases44. Therefore, clinicians might prefer to prescribe ziprasidone for patients with both schizophrenia and metabolic abnormalities. This might explain why the ziprasidone group in our analysis has a higher proportion of metabolic syndrome and metabolic abnormalities at baseline. However, the lurasidone group was excluded due to the limited sample size (only two patients), and the amisulpride and aripiprazole groups only included 13 and 10 patients, lowering the reliability and representativeness of their influences on outcome changes. Further real-world studies with larger sample size or randomized clinical trials in the context of clinical environments are warranted to answer of the relationships of prescribed other low-risk SGAs and metabolic outcomes.

The strengths of our study included the focus on patients with both schizophrenia and metabolic abnormalities, the application of PSM, the comprehensive measurement of outcomes (proportion of metabolic syndrome, the number of metabolic abnormalities and the level of metabolic parameters). The current study also has several limitations. First, although 2189 patients with schizophrenia were initially recruited, only 764 met the inclusion criteria. After PSM, only 223 matched pairs were in the final analysis, lowering the statistical power and damaging the representativeness of the results. In addition, only 423 patients prescribed one SGA, with 2, 13 and 10 prescribed lurasidone, amisulpride and aripiprazole, precluding us to compare the associations of these low-risk antipsychotics. Second, the decision to add or switch SGAs with a low metabolic risk should consider the clinical environment and patients’ preferences, therefore the treatment was not randomly assigned. Although PSM has been performed to balance baseline characteristics between groups, confounding bias may exist. Third, the 6-month follow-up period was relatively short, and the long-term effectiveness and safety of adding or switching low-risk SGAs was unable to observe. In addition, previous evidence has shown that aripiprazole could induce weight gain over longer treatment durations45. Therefore, our classification of SGAs by metabolic risk might cause bias. Fourth, only five high/intermediate-risk SGAs and three low-risk SGAs were included, and whether and to what extent other low-risk SGAs would influence the metabolic outcomes for patients with both schizophrenia and metabolic abnormalities need further verification. Fifth, due to the small sample size, we did not examine the differences across sex, age, and duration of schizophrenia, although previous research has identified these variables as important indicators of metabolic abnormalities46,47. In addition, information on the adherence to SGAs and adverse effects was unavailable, therefore our results may be biased by low drug adherence. Sixth, participants in the high/intermediate SGAs group might fail to switch or add SGAs with low metabolic risk due to many unknown reasons (i.e., patients were too unstable to switch), which PSM cannot fully resolve. Seventh, the information on diet and physical activity was unavailable in our study, which have been widely regard as risk factors of metabolic outcomes.

Conclusion

Adding or switching to SGAs with a low metabolic risk for patients with schizophrenia was associated with alleviating metabolic syndrome and metabolic abnormalities for six months, especially in weight glucose, and triglyceride. These associations were more pronounced in those with no comorbid physical conditions, lower BMI and shorter duration. Adding or switching to low-risk antipsychotics can be an appropriate secondary prevention strategy for patients with both schizophrenia and metabolic abnormalities.

Ethics approval and informed consent

Ethics approvals were received from the Institution Review Board at Zhejiang Provincial People’s Hospital (KY2022027). All informed participants provided written consent.