Comparison of Long-Acting Injectable Antipsychotics With Oral Antipsychotics and Suicide and All-Cause Mortality in Patients With Newly Diagnosed Schizophrenia

Key Points Question Is use of long-acting injectable antipsychotics (LAIs) in patients with newly diagnosed schizophrenia associated with decreased mortality? Findings This cohort study included 2614 patients with schizophrenia who switched to LAIs, and 2614 propensity-matched patients who continued receiving oral antipsychotics (OAPs) of the same compounds. Patients who switched to LAIs had lower mortality and fewer suicide attempts compared with patients who continued receiving the corresponding OAPs, and patients who switched to LAIs within the first 2 years of OAP initiation exhibited a 47% reduction in the risk of suicide mortality compared with patients who continued receiving OAPs. Meaning These findings suggest that LAI use in patients with newly diagnosed schizophrenia should be encouraged because of its association with lower premature mortality.


Introduction
Schizophrenia is generally considered to be among the most severe psychiatric disorders because of the excessive mortality associated with it. It has consistently been associated with a life expectancy 10 to 25 years shorter than that of the general population. 1,2 In a meta-analysis, Brown et al 3 suggested that schizophrenia is associated with a large increase in suicide mortality and a moderately increased risk of natural-cause mortality. Therefore, research on practical methods for reducing these risks is highly warranted.
The lifetime risk of suicide in patients with schizophrenia is approximately 5%. 4 The risk is highest during the early stage of the illness or the first episode. 5 The suicide rate for patients with schizophrenia spectrum disorders is more than 20 times higher than that for the general population. 6 Although suicide is a complex public health challenge, suicide is typically preventable with timely and evidence-based interventions. Risk factors for suicide in this population include being young, male, and highly educated. Illness-related risk factors include prior suicide attempts, depressive symptoms, active positive symptoms, and comorbid substance misuse. 4,7 Effective treatment delivery and adherence are vital for suicide prevention in schizophrenia. 4,7 Compared with oral antipsychotics (OAPs), long-acting injectable antipsychotics (LAIs) were reported to improve treatment adherence in patients with schizophrenia. 8,9 However, studies of LAIs in protecting against suicide are limited; in particular, the association of LAIs with the course of the illness, if administered at an early phase, remains unclear.
The objective of the current study, therefore, was to investigate associations of LAIs and their equivalent OAPs on risks of all-cause, natural-cause, and suicide mortality in patients with newly diagnosed schizophrenia. We also explored whether patients benefited from the early use of LAIs.

Data Source
We conducted a population-based cohort study using data from the Taiwan National Health Insurance (NHI) Research Database (NHIRD) and the Taiwan Death Registry (TDR) database. The NHIRD is derived from Taiwan's single-payer compulsory NHI program, which covers up to 99% of the 23 million people in Taiwan. The NHIRD includes all medical claims data on disease diagnoses, procedures, drug prescriptions, demographic characteristics, and enrollment profiles of all NHI beneficiaries. 10 In addition, we linked the NHIRD to the TDR database to ascertain the vital status, date of death, and cause of death of patients. The accuracy of cause-of-death coding in the TDR database was previously validated. 11 This study was reviewed and approved by the institutional review board of Taipei Medical University. A waiver of informed consent was granted because the patient information in the national claims data from the NHIRD was deidentified before analysis. All researchers signed an agreement guaranteeing patient confidentiality before using the database. This study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies. 12 the base cohort. The base cohort entry date was defined as the date of the first antipsychotic prescription.

Study Cohort
In our study, the LAI group was defined as patients in the base cohort who initiated an OAP prescription and then switched to LAIs during subsequent treatment and were prescribed LAIs at least 4 times within 1 year. 13,14 In total, 35 182 patients (25%) switched to LAIs, and 13 826 (10%) were prescribed LAIs at least 4 times within 1 year. For each patient who received LAIs, we selected a propensity-matched reference patient from our base cohort who received an OAP of the same compound and continued receiving OAPs throughout. Using a prevalent new-user design, 15 reference patients were selected from OAP initiators who spent the same duration of time in the base cohort as did exposed patients (defined as the period from the first prescription of OAPs to the first prescription of LAI) and continued receiving OAPs when exposed patients switched to LAIs. Accordingly, we randomly selected up to 1 reference patient in each exposure set who had matched OAPs of the same compounds received, age at cohort entry, sex, calendar year of cohort entry, psychiatric comorbidities 1 year before the index date, and psychiatric hospitalization (yes or no) 6 months before the index date with cases, using a time-conditioned propensity score. The index date was defined as the date of the first LAI prescription. All patients were followed up until switching the antipsychotic administration route, death, or the end of the study (December 31, 2018), whichever occurred first. The study flowchart and timeline are presented in eFigure 1 and eFigure 2 in the Supplement.

Antipsychotic Exposure
Information on antipsychotics use was obtained from ambulatory and inpatient prescription claims data using the Anatomical Therapeutic Chemical Classification System code N05A (antipsychotics) but excluding N05AN (lithium). 16 The antipsychotics and Anatomical Therapeutic Chemical codes used for the definition are shown in eTable 1 in the Supplement. Patients were stratified into LAI and OAP groups according to the study criteria. LAIs were further categorized into flupentixol, haloperidol, olanzapine, paliperidone, and risperidone to explore the associations of different drugs with the outcomes. To investigate the association of the early use of LAIs, we grouped patients using LAIs into those switching to LAIs within 2 years and those switching after 2 years, according to a literature review. [17][18][19][20]

Statistical Analysis
Descriptive statistics were used to summarize patient characteristics in the LAI and OAP groups.
Potential imbalances among covariates after matching were assessed using standardized mean differences. All-cause, natural-cause, and suicide mortalities per 100 000 person-years were independently estimated for the LAI and OAP groups. Conditional Cox regressions were used to estimate the risk of death. 24 The Fine and Gray 25 method was adapted to estimate the hazard of natural-cause and suicide deaths, considering competing risks from other causes of death. All Cox models included the CCI score within 1 year before the index date, history of suicide attempts, and number of psychiatric ED visits within 6 months before the index date in the model for adjustment.
We used a Kolmogorov-type supremum test based on 1000 simulated residual patterns to test the validity of the proportional hazards assumption. 26

Patient Characteristics
The database extracted for the current analysis comprised 5228 patients with newly diagnosed  [23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38][39] years) patients in our study (1333 male patients [51.0%] in each group). The mean (SD) interval from the first LAI prescription to the fourth LAI prescription was 109 (69.3) days. Compared with the OAP group, the LAI group was more likely to have a history of suicide attempts and psychiatric ED visits 6 months before the index date, which implied that the LAI group consisted of patients with clinically more-severe disease ( Table 1). The covariates mentioned were included in the matching and were evenly distributed between the 2 groups, except for the CCI score, history of suicide attempts, and number of psychiatric ED visits. Thus, these 3 covariates were not included in the matching.

Suicide Attempts
During the follow-up period, patients who switched to LAIs had fewer hospital visits for suicide attempts than their matched OAP controls (3.5 vs 4.2 visits) ( Table 3). The negative binomial regression results indicated that patients who switched to LAIs exhibited a 28% lower risk of suicide attempts than patients who received only OAPs (incidence rate ratio, 0.72; 95% CI, 0.55-0.93).

Benefit of Early Use of LAIs
We examined the risk of mortality associated with the early use of LAIs (ie, within 2 years) compared with the matched OAP group (

Discussion
In this study, the use of LAIs was associated with significantly lower risks of all-and natural-cause mortality and suicide attempts than was the use of OAPs after adjusting for covariates. With regard to the benefits of early use of LAIs, patients who switched to LAIs within the first 2 years of OAP initiation had a 47% decrease in the suicide mortality risk compared with those who continued receiving OAPs.
Here, the LAI group had a 34% lower all-cause and a 37% lower natural-cause mortality risk than did the OAP group. These results are in line with those of previous large observational cohort studies and meta-analyses of randomized clinical trials, in which receiving LAIs was associated with lower mortality than was receiving OAPs. 9,28,29 Brown et al 30 suggested that some of the excess mortality of schizophrenia could be lessened by reducing patient exposure to environmental risk factors and improving disease, mood disturbance, and psychosis management. Because LAIs were associated with lower risks of rehospitalization and antipsychotic treatment discontinuation than were OAPs, 9,14,29 better outcomes in all-cause and natural-cause mortality might be attributable to improved medication adherence and less disease fluctuation and relapse.
With regard to associations of individual antipsychotics with mortality, the LAI forms of paliperidone, haloperidol, and risperidone were associated with lower risks of all-cause mortality compared with their corresponding OAPs. Notably, LAI flupentixol and LAI olanzapine were not associated with significant benefits for all-cause mortality. These findings are consistent with those of Taipale et al. 28 However, studies further assessing the efficacy of individual antipsychotics against mortality are needed.
The lifetime risk of suicide in patients with schizophrenia is approximately 5%, 4 and suicide is another leading contributor to excessive premature mortality in this population. 6,30 The risk is highest among patients during the early stage of their illness or their first episode. 5 Most suicide attempts occur within 2 years after the first episode of psychosis, 31,32 and suicide-related mortality is higher among individuals recently diagnosed as having schizophrenia (ie, Յ5 years from diagnosis). 33 The higher risk of suicide compared with later stages of schizophrenia might be explained by the critical period hypothesis, which states that the primary clinical and psychosocial deterioration associated with schizophrenia occurs within the first 5 years, the critical period. 34,35 Within this period, biological, psychological, and psychosocial influences are developing and have maximum plasticity to adequate interventions. However, relapse is also common in this period because of low levels of insight and nonadherence to medication. 36 Although suicide is a complex and multifactorial phenomenon, suicides are typically preventable with timely, evidence-based interventions. Hence, active early intervention and even the early application of LAIs to reduce risks of relapse, suicide attempts, and suicide mortality in the early stage have been proposed. [36][37][38][39] Our findings suggest that LAI users had a lower suicide attempt risk and that patients who switched to LAIs within 2 years of OAPs treatment had decreased suicide mortality compared with  those who continued receiving OAPs. These results are consistent with a previous study 38 that revealed that the use of second-generation LAIs for recent-onset schizophrenia reduced suicidal ideation compared with chronic schizophrenia. However, we did not observe a decrease in suicide mortality among patients who switched to LAIs more than 2 years after OAP treatment. This might have been due to the neurological processes underlying schizophrenia having been ongoing for many years before the first episode. 40 Therefore, with later LAI use, their protective effects might not be sufficient to compensate for the cognitive and psychosocial deterioration that occurs in the first few years of the illness.
Clinically, most psychiatrists use LAIs with a conservative attitiude, 41,42 and the reasons for this attitude are generally not well supported by current scientific evidence. 41 This attitude was also observed in our study; only 10% of our base cohort was prescribed LAIs at least 4 times within 1 year.
Heres et al 43  early-stage schizophrenia in the majority of guidelines. 14,44,45 In our findings, the LAI group had significantly lower risks of all-cause and natural-cause mortality, and those who switched to LAIs in the early stage (Յ2 years) had a 47% decreased risk in suicide mortality as well. Given that inadequate medication adherence is common among patients with newly diagnosed schizophrenia, more active consideration of LAIs in this stage for better longterm outcomes should be encouraged, particularly for those who have already exhibited poor adherence attitudes. 37,39,[45][46][47][48] Additional studies are also needed to clarify long-term adverse effects specific to LAIs use and identify specific psychotherapeutic and psychosocial interventions that may offer other benefits to patients in the early stage.

Strengths and Limitations
This study's major strength is its population-based design that used a prevalent new-user design to examine long-term mortality, for up to 16 years, as our primary outcome in patients with newly diagnosed schizophrenia. This design enabled head-to-head comparisons between LAIs and their corresponding OAPs by enrolling patients and considering the disease duration, age, sex, and psychiatric comorbidities at the baseline.
Several limitations should be addressed. First, all observational studies have residual confounding. However, we made our best effort to minimize this potential bias by stringently selecting our matched cohort using the duration of OAPs treatment, year of first antipsychotic treatment, and psychotic comorbidities at the baseline. Second, lifestyle factors, socioeconomic status, social support, and the personal preferences of psychiatrists were not included in the study because of the NHIRD's limitations. Third, the CCI score, history of suicide attempts, and number of psychiatric ED visits were included in the Cox hazard model but not in the matching. Fourth, an Figure