This real-world study yielded three important findings regarding individuals with schizophrenia in China. First, more than one in ten patients had reported a history of self-harm on admission. Second, a lifetime history of self-harm was associated with a higher possibility of clozapine uses on admission. Third, a lifetime history of suicidal behavior increased the possibility of initiating clozapine treatment during hospitalization or within 56 days after admission.

In our study, four-fifths of the patients used risperidone, olanzapine, and aripiprazole on admission. Clozapine use was uncommon (1.2% on admission and 2.7% started during hospitalization) for the treatment of early-stage schizophrenia at our sample site. Another study on drug-naïve first-episode schizophrenia spectrum disorders in a psychiatric hospital in Hunan, China reported a higher clozapine use rate (7.5%), which was 4.5% in tertiary hospitals and 11% in non-tertiary hospitals [29]. Their clozapine use rate in tertiary hospitals was similar in our study (also a tertiary hospital). This explanation may be because psychiatrists in high-level hospitals adherence more to China’s national clinical guidelines [29]. In China, clozapine is indicated only for treatment-resistant schizophrenia [14].

Our study found that psychiatrists preferred to clozapine or changed prescription to it sooner when an admitted patient had a lifetime history of suicidal behavior. The first reason may be the psychiatrists acknowledged that, in the United States, clozapine is the only antipsychotic with an indication for treating suicide risk [12]. International best practices influence clinical decision making. The second reason may be a history of suicidal behavior increased their clinical prediction of the risk of treatment-resistant schizophrenia, an indication for clozapine use. Considering the small number of patients using clozapine, there may be potential chance reasons for the increased likelihood of switching to clozapine earlier among those with a history of suicidal behavior. Future research should test these findings using a larger sample size.

None of the patient used LAI antipsychotics, although LAI antipsychotics (risperidone, paliperidone, and perphenazine) were available at research site. Our sample consisted of inpatients with a short duration of schizophrenia. Then, the evidence of LAI antipsychotics and subsequent suicides has been rare [30].

Suicide prevention is a challenge in the treatment of schizophrenia in psychiatric hospitals. A previous study reported that the majority (64/79) of suicidal events in a Chinese psychiatric hospital occurred from those with a diagnosis of schizophrenia [31]. Our study provides clinicians with information on the prevalence of self-harm among individuals with early-stage schizophrenia, suggesting the need to prevent their risk of self-harm and suicide. The findings about antipsychotic use upon admission and during hospitalization suggest the use of clozapine is not as popular as risperidone, olanzapine, and aripiprazole. The guidelines of treatment for schizophrenia should emphasize the underutilization among patients with a history of self-harm. In fact, clozapine is underutilized in many countries [32], although research suggests that clozapine reduces self-harm [7,8,9] and mortality compared with other antipsychotics [10, 33]. Underutilization may become a barrier to recovery from schizophrenia and high suicide risk.

In addition, our results showed that clozapine is utilized sooner for those patients with a history of suicidal behavior, who may benefit in reducing suicide risk from clozapine use. This survey on clinical practice can help psychiatrists know what other psychiatrists did when treating schizophrenia patients and not to be too cautious about clozapine. Then, clinicians should be trained to become more familiar with managing the risks and side effects of clozapine and its efficacy in preventing suicide and increasing life expectancy [34]. This training could change the situation of “clozaphobia” and improve the prescribing of clozapine for those really need it. In clinical practice, psychiatrists can choose tools such as multidimensional clozapine side-effect management [33] and therapeutic drug monitoring are essential tools, to increase clozapine safety. Reducing the clozapine dose is necessary for patients with reduced CYP2D6 (cytochrome P450 2D6; CYP2D6 poor metabolizers) activity. It is also important to regularly monitor the ANC on a regular basis for at least 2 years [35]. Comprehensive care for other side effects, such as cardiovascular risk, metabolic issues, and clozapine-induced constipation, would also optimize the treatment. One obstacle is implementing the implementation of evidence-based strategies.

This real-world study used EMR’s prescription data to minimize information bias regarding antipsychotic use. Moreover, before analyzing the clozapine prescriptions in the cohort, we used blood test data and diagnoses of leukopenia or thrombocytopenia from the EMR to exclude clozapine contraindications. We also reported the short-term prescription switch that will help other practitioners choose an earlier use of clozapine.

This study has several limitations. First, the findings may be confounded by unmeasured factors, such as psychotic symptom severity, personal choice, previous antipsychotic use, antidepressants, ECT therapy, and self-harm during hospitalizations. Moreover, missing information on antipsychotic use before hospitalization and antipsychotic combination practice made it difficult to evaluate whether patients received a full course of treatment. Thus, we could not determine whether the patients who received clozapine in this study aligned with or violated the label.

Second, recall bias, stigma, and patient’ incapability may lead to reduced self-harm disclosure. The absence of a description of self-harm or suicidal behavior in medical records does not always indicate the absence of such behaviors. However, because our data were obtained from the inpatient department, the history of self-harm behavior was checked repeatedly by the entire clinical team, including the consultant psychiatrist, attending doctor, resident doctor, and nurses. Moreover, the bias in suicidal reporting may not have influenced our finding of an association between self-reported history of self-harm and psychiatrists’ decisions on antipsychotics. If the clinical team did not collect information on the history of suicide, psychiatrists would not consider using clozapine to reduce suicide risk.

Third, the data were collected from only one site, limiting the generalizability of the findings. However, we did not exclude patients who had received short-term schizophrenia treatment at other hospitals. Prescriptions at admission also reflected the practice in the previous hospital.

Fourth, we only included inpatients. Hospital psychiatry care is common among individuals with acute-phase schizophrenia in China. Self-harm intention or behavior is an indication for hospitalization. This sample have overestimated the prevalence of self-harm and clozapine use among all patients with early-stage schizophrenia.

Fifth, the sample size of the Cox regression analysis was limited, and the confidence intervals were wide for some estimates.

Finally, we directly used the diagnosis of schizophrenia from the EMR. To help psychiatrists differentiate schizophrenia diagnosis from other psychoses caused by organic/toxic causes, every inpatient with psychosis symptoms in the research site was routinely required to undergo brain imaging, electroencephalography, and neuropsychological assessments. However, lumbar puncture or urinalysis was performed only when necessary.