Association of antipsychotic formulations with sudden cardiac death in patients with schizophrenia: A nationwide population-based case–control study

.


Highlight
• Question Are antipsychotic formulations associated with sudden cardiac death (SCD) in patients with schizophrenia?
• Findings In patients with schizophrenia, the risk of SCD with different antipsychotic formulations is ranked from high to low as follows: long-acting injectable antipsychotic (LAI) and oral-daily antipsychotic (OAP) combination, LAI monotherapy, OAP monotherapy, and nonantipsychotic use within the observation period.In addition, cardiovascular comorbidities are associated with an increased risk of SCD.
• Meaning Physicians should obtain the history of cardiovascular disease before prescribing LAIs to patients with schizophrenia.They should also minimize prescribing combined antipsychotics and closely monitor their patients' cardiovascular condition after prescribing LAIs.

Introduction
Schizophrenia is a debilitating psychotic disorder that affects approximately 1% of the world's population (Freedman, 2003).It often leads to premature death, shortening life expectancy by 10 to 25 years (Crump et al., 2013).Compared with the general population, patients with schizophrenia face a threefold higher risk of sudden unexpected death (Appleby et al., 2000;Ruschena et al., 1998).Among the leading causes of this premature mortality are suicide, accidents, and sudden cardiac death (SCD) (Crump et al., 2013;Saha et al., 2007;Sweeting et al., 2013).
Antipsychotic medications form the cornerstone of treatment for schizophrenia (Lally and MacCabe, 2015;Zhu et al., 2019).These can be categorized into first-generation (typical) and second-generation (atypical) antipsychotics on the basis of their pharmacodynamics.In addition, they can be classified as oral-daily antipsychotics (OAPs), oral-weekly antipsychotics, short-acting injectable antipsychotics, and long-acting injectable antipsychotics (LAIs) on the basis of their formulation.Consistent, long-term pharmacotherapy is crucial for effective treatment, symptom management, and relapse prevention in schizophrenia (American Psychiatric Association, 2006;Zhu et al., 2019).However, over half of patients with schizophrenia exhibit poor or no adherence to medication (Byerly et al., 2007;Chan et al., 2017;Lacro et al., 2002;Velligan et al., 2006).The COVID-19 pandemic prompted mental health professionals to seek more effective methods for preventing relapse and reducing unplanned hospitalizations in these patients.Such methods are imperative because patients with severe psychotic symptoms that are not appropriately managed may place themselves at risk of infection and potentially contribute to the spread of COVID-19, which can undermine infection control efforts (MacLaurin et al., 2021).LAIs have emerged as an optimal pharmacotherapeutic regimen to prevent relapse in patients with schizophrenia, particularly young patients who struggle with medication adherence (Miron et al., 2022).
In patients with schizophrenia, antipsychotic medications are associated with an increased risk of premature death from SCD. Patients on either first-or second-generation antipsychotics typically face a similar, dose-related increased risk of SCD (Jones et al., 2013;Ray et al., 2001;Ray et al., 2009;R. Girardin et al., 2013;Straus et al., 2004;Wu et al., 2015).Although the occurrence of antipsychoticinduced SCD is rare, the potential for fatal adverse effects may lead to restrictions on prescribing certain antipsychotics, particularly LAIs, to patients with schizophrenia (Zhu et al., 2019).However, evidence regarding the association between antipsychotic formulations, such as OAPs, LAIs, and their combination, and the risk of SCD in patients with schizophrenia is scarce.Therefore, this nested casecontrol study investigated the association between different antipsychotic formulations and the risk of SCD in patients with newly diagnosed schizophrenia by using a nationwide population-based claims database.

Data sources and ethics statement
This study used multiple-cause-of-death (MCOD) data from Taiwan and deidentified nationwide,

Study design, case selection, and case-control matching
This study adopted a nationwide, population-based, nested case-control design.We defined cohort patients as individuals with a new diagnosis of schizophrenia from January 1, 2012, to December 31, 2018, according to the NHIRD.Cases in this study were patients who experienced SCD between 2012 and 2020.Controls were defined as individuals with schizophrenia who did not experience SCD; they were selected randomly and matched to cases at a 1:4 ratio by age, sex, and year of new schizophrenia diagnosis, as illustrated in Fig. 1.
We identified patients with schizophrenia by using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 295 and the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) codes F20 and F25.To improve diagnostic validity, we included the inpatients with schizophrenia diagnoses from psychiatric departments and outpatients with schizophrenia diagnoses who received antipsychotic prescriptions for more than 3 months within 180 days after a new diagnosis.We excluded patients with missing data on sex or age, those younger than 20 years, and older than 60 years, considering the rarity of SCD in younger individuals (Liberthson, 1996) and differing aetiologies in older individuals.Ultimately, 28,471 individuals were included in the study.
The primary event of interest was SCD occurring after the exposure of different antipsychotic formulations.SCD was defined as immediate death with an underlying cardiac disease cause, identified using cardiac death codes (ICD-9-CM and ICD-10-CM codes) from the NHIRD and MCOD data (Chung et al., 2010) (see Table S2).The date of each SCD event was the index date for cases.Four controls were matched for each case by age, sex, and the year of new schizophrenia diagnosis from the same cohorts without SCD.Controls were selected on the basis of their index date, which corresponded to each case's index date, and on the basis of whether they were alive at that time.

Antipsychotic exposure definition
This study analysed antipsychotic usage on the basis of outpatient and inpatient prescription data.We used the Anatomical Therapeutic Chemical classification system code N05A to identify antipsychotic prescriptions, excluding N05AN (lithium; see Table S3).Antipsychotic use was evaluated by examining antipsychotic exposure in the 60-day observational period before the index date, which is longer than that in a previous study that evaluated OAP use with a 30-day observational period for carryover effects (Straus et al., 2004).This extended observational period was used to account for any potential, yet unidentified, carryover effects associated with LAIs and thereby ensure a comprehensive analysis of antipsychotic use in both cases and controls.Antipsychotic formulations were divided into three types: OAP monotherapy, LAI monotherapy, and LAI and OAP combination.Patients were classified into one of the following three groups of the basis of their antipsychotic exposure during the 60-day observation period: patients who had ever used LAIs, those using OAP monotherapy, and those who did not use any antipsychotics.Patients who had ever used LAIs during the observational period were further divided into two subgroups: those who had used LAI monotherapy and those who had used LAI and OAP combination.Nonantipsychotic use was defined as no use of any antipsychotic medication during the observational period.

Covariates
We collected demographic data, including age, sex, and the year of new schizophrenia diagnosis.
Comedications included digoxin and drugs known to increase the risk of torsade de pointes (TdP) (Glassman and Bigger, 2001;Suvisaari et al., 2010) (see Table S3).The number of psychiatric ward admissions during the 3 years before the index date served as a measure of psychiatric health.

Statistical analysis
This study employed inverse probability weighting (IPW) to ensure comparability between cases and controls.Propensity scores were calculated using age; sex; year of new schizophrenia diagnosis; severity of comorbidities; and presence of arrhythmia, hypertension, epilepsy, depression or other mood disorders, ischemic cerebrovascular disease, diabetes mellitus, heart failure, alcohol abuse or alcoholism, cardiovascular disease, obesity, use of digoxin, use of drugs with a known risk of TdP in the year before the index date, and the number of hospitalizations.We then applied IPW, assigning weights to cases and controls inversely proportional to their probability.Weights were calculated as 1/propensity score for cases and 1/ (1 − propensity score) for controls (Chesnaye et al., 2012).The standardized mean difference (SMD) was used to evaluate the balance of covariates, with an absolute SMD below 0.1 indicating a balanced distribution (Zhang et al., 2019).
We used logistic regression with and without IPW adjustment to calculate odds ratios (OR) for different antipsychotic formulation groups among the cases and controls.The logistic regression variables (without IPW) mirrored those used in calculating propensity scores.A two-sided p < 0.05 was set as the threshold for statistical significance.Although relevant studies have identified a 30-day carryover effect of OAP monotherapy on SCD, the potential for a similar effect in LAIs remains to be established.To ensure robust results, we assessed antipsychotic exposure over intervals of 30, 90, and 180 days before the index date.All statistical analyses were conducted using SAS software version 9•4 (SAS Institute, Cary, NC, USA).

Patient characteristics
Over the study period, 28,471 patients received a new diagnosis of schizophrenia and were therefore included in the cohort.From these patients, 366 who experienced SCD were classified as cases, whereas 1464 who did not experience SCD were designated as controls (see Fig. 1).The mean (SD) age for both groups was 47.0 years with an SD of 9.1 years; the majority of the patients were men (66.0%), and the patients exhibited a similar distribution of diagnosis years (see Table 1).
Before IPW and adjustment for propensity scores, significant differences were observed between the cases and controls in terms of monthly income, CCI score, presence of all comorbidities (except for hyperlipidaemia and obesity), comedications, and number of hospitalizations (see Table 1).After IPW was applied, these baseline characteristics were balanced, with absolute SMDs lower than 0•1 noted for all variables, with the exception of comedication with digoxin, which had a slightly higher value (Table 1).Upon adjustment for propensity scores, significant differences were observed only in the cardiovascular comorbidities between the two groups (adjusted OR [aOR] = 11.15,95% CI 4.60-27.03,p < 0.001).No significant differences were observed in the number of hospitalizations (see Table 1 and   Table S1).
When IPW was applied, the results from the sensitivity analysis aligned with those of the primary analysis for the 60-day observational period, for logistic regression with and without IPW.Notably, IPW analysis revealed that the use of LAI monotherapy was associated with a significantly higher risk of SCD at 30 days (OR = 1.39, 95% CI 1.04-1.87,p = 0.029), 90 days (OR = 1.47, 95% CI 1.08-1.99,p = 0.014), and 180 days (OR = 1.50, 95% CI 1.08-2.09,p = 0.017) that was the use of OAP monotherapy.

Discussion
To our knowledge, this is the first nationwide case-control study to investigate the relationship between antipsychotic formulations and the risk of SCD in patients with a new diagnosis of schizophrenia.Our findings reveal the following hierarchy of SCD risk associated with antipsychotic formulations (listed from highest to lowest risk): LAI and OAP combination, LAI monotherapy, OAP monotherapy, and no antipsychotic use within a 60-day observational window.Adjustments for known confounders and IPW indicated that the LAI and OAP combination was associated with an approximately doubled risk of SCD.Sensitivity analyses across the observational periods of 30, 90, and 180 days confirmed the consistency of the findings for these periods with those of the 60-day observational period.
In addition to antipsychotic formulations, cardiovascular comorbidities were identified as a significant risk factor for SCD in patients with schizophrenia.
After the COVID-19 pandemic, the focus of several mental health professionals shifted toward the use of LAIs in patients with schizophrenia, particularly for those with poor or no medication adherence.This shift was due to the effectiveness of LAIs in reducing relapse and unplanned hospitalizations.
However, evidence regarding the cardiac safety of LAIs in patients with schizophrenia is sparse.Studies have indicated that first-generation antipsychotics significantly elevate SCD risk compared with that when no antipsychotics are used (Hou et al., 2015;Ray et al., 2001;Straus et al., 2004).Other studies from the United States (Miron et al., 2022), United Kingdom (Jones et al., 2013), Switzerland (R. Girardin et al., 2013), and Taiwan (Hou et al., 2015) have revealed firstand second-generation antipsychotics to be associated with a dose-related increase in SCD risk.However, these studies have often not employed comprehensive analysis of different antipsychotic formulations, such as OAPs and LAIs, and included study populations with only a small proportion of patients with schizophrenia (8% to 20%) (Jones et al., 2013;Ray et al., 2001;Ray et al., 2009;R. Girardin et al., 2013;Straus et al., 2004;Wu et al., 2015).To assess the cardiac safety of LAIs in patients with schizophrenia, we investigated the correlations between various antipsychotic formulations and the risk of SCD in patients with newly diagnosed schizophrenia.Our results indicate that patients with schizophrenia who had used LAIs at any time faced a higher risk of SCD than did those who solely used OAPs or who did not use antipsychotics.
Notably, the risk was most pronounced in patients prescribed a combination of LAIs and OAPs.These findings highlight the importance of carefully considering the appropriateness of LAIs as a treatment option for patients with schizophrenia and the need for doctors to exercise caution when prescribing a combination of LAIs and OAPs for this patient population.
We investigated potential confounders other than cardiovascular comorbidities to determine their association with SCD in patients with schizophrenia.A key finding was pre-existing cardiovascular disease being a significant risk factor for SCD in these patients.Therefore, before prescribing LAIs to patients with schizophrenia, physicians should identify cardiovascular comorbidities and conduct initial cardiovascular screening using electrocardiography and biochemical analyses.After prescribing LAIs, physicians should closely monitor their patients for the emergence of any cardiovascular disease.This study did not evaluate the mechanisms by which different antipsychotic formulations induced SCD.Multiple probable mechanisms have been proposed to underlie antipsychotic-induced SCD, including ventricle repolarization abnormalities, genetic predisposition , autonomic effects, other acute cardiotoxicity, inhibition of other ion channels, and pharmacokinetic and pharmacodynamic interactions (Abdelmawla and Mitchell, 2006;Buckley et al., 2000;Ray et al., 2009;Sheehan et al., 2012;Sicouri and Antzelevitch, 2018;Straus et al., 2004).Antipsychotic medications may increase the risk of serious ventricular arrhythmias such as TdP, probably QT interval prolongation through blockade of potassium ion channels (Ray et al., 2009;Straus et al., 2004).Furthermore, a genetic mechanism by which antipsychotics might induce SCD relates to the gene mutations encoding potassium ion channel proteins (i.e., SCN5A, KCNQ1, and KCNH2) (Sicouri and Antzelevitch, 2018).Specifically, polypharmacy and high-dose antipsychotics are widely considered to be an important and the clearest evidence for an elevated SCD risk related to drug-drug interactions in pharmacodynamics and pharmacokinetics (Abdelmawla and Mitchell, 2006).Pharmacokinetic mechanism has proposed to have higher SCD risk of LAIs than that of OAPs.Compared with OAPs, larger fluctuations in peak-to-through concentrations of LAIs may negatively impact clinical response and adverse effects (Correll et al., 2019;Sheehan et al., 2012).Notably, peak concentration may be also an indicator of severity of adverse effects including SCD (Correll et al., 2019;Sheehan et al., 2012).
The primary strength of this study lies in its large sample size, which represents nearly the entire population of Taiwan.This extensive data set enabled detailed investigation of the association between antipsychotic formulations and the risk of SCD in patients with newly diagnosed schizophrenia.Our results provide valuable insights into the relationship between antipsychotic formulations and the risk of SCD that are particularly relevant in the context of extensive use of LAIs after the COVID-19 pandemic.
Another key strength of this study is its comprehensive establishment of a temporal association between antipsychotic formulations and the occurrence of SCD.To increase the sensitivity of our analysis of this relationship, we evaluated exposure to antipsychotic drugs over different observational periods.We also used propensity score matching and IPW to mitigate biases potentially arising from suboptimal specificity of models for estimating propensity scores or outcome regressions (Chan et al., 2017).Although the study was retrospective, the use of a nested case-control approach within a nationwide, population-based database minimized the likelihood of recall and selection biases.
Despite these strengths, this study has certain limitations.First, we could not obtain data on smoking habits from the database.This omission is notable because smoking is a potential SCD risk factor and the prevalence of smoking is high among patients with schizophrenia (Kannel et al., 1975;Ray et al., 2001).
Second, because of the nature of the healthcare database used in this study, the effect of drug-naïve patients with schizophrenia on our analysis may have been minimal.To approximate the conditions of drug-naïve patients with schizophrenia, we assessed the risk of SCD among treated patients with schizophrenia who had not used antipsychotics during various observation periods.Third, medication adherence among patients with schizophrenia was not evaluated because data on drug use, particularly OAP use, could only be obtained from medical claims data.In addition, the prevalence of schizophrenia and comorbidities such as hypertension, hyperlipemia, cardiovascular disease, and obesity, may have been underestimated because patients experiencing mild symptoms might not seek medical services.Fourth, we could not include data on dietary and other potential confounders for SCD, such as disease severity, biochemistry, electrocardiography results (Hou et al., 2015), and unhealthy lifestyle factors such as alcohol use, sweetened beverage consumption, and inadequate exercise.Finally, the case-crossover study design was performed and we also compared to the results with our case-control study design.In the case-crossover design, three durations (15-day, 30-day, and 60-day) were chosen a priori to assess acute to mid-term risk for primary case and control time-period comparisons.In this design, the risk of SCD with different antipsychotic formulations was ranked from high to low as follows: LAI monotherapy, OAP monotherapy, and non-antipsychotic use (see Table S6).These findings had consistent tendency between both designs despite nonsignificant difference in the case-crossover design.One possible explanation for the nonsignificant difference is that the sample size of study cases was small.
In conclusion, the LAI and OAP combination in patients with schizophrenia is associated with a higher risk of SCD, particularly when cardiovascular comorbidities are considered, than are other antipsychotic formulations.Overall, our findings provide valuable insights into the correlation between LAIs and the risk of SCD and contribute valuable information that can assist in the treatment decision-making process for patients with schizophrenia who are prescribed LAIs.Providing such treatment involves balancing the benefits of LAI treatment in patients with poor and no medication adherence against the associated risks.
Notably, the approximately twofold increase in SCD risk observed in our study, although significant, is not conclusively causally related to the use of OAP monotherapy.Considering our findings, we advocate for comprehensive cardiovascular disease history assessment before LAIs are prescribed to patients with schizophrenia.Physicians should also minimize prescribing combined antipsychotic regimens.Close monitoring of the cardiovascular health of patients after a prescription of LAIs is crucial to ensure early intervention and support when required.The cut-off point of SMD was set as 0.1; the absolute SMD<0.1 would be viewed as the balance of the covariates.
population-level claims-based data sourced from the National Health Insurance Research Database (NHIRD) of Taiwan for the period spanning 2011 to 2020.The NHIRD is a comprehensive, real-world database that provides insights into the clinical profiles of patients enrolled in the National Health Insurance program, which was established on March 1, 1995.This program covers approximately 99% of the Taiwanese population of approximately 23 million individuals (National Health Insurance Research Database, 2018).The data in the NHIRD are managed by the Health and Welfare Data Science Centre of Taiwan.This study was approved by the Institutional Review Board of Dalin Tzu Chi Hospital, Buddhist Tzu Chi Medical Foundation, Chiayi, Taiwan (ethics approval number B11001023).Because the data used in this study were deidentified, the requirement for informed consent was waived.The study was conducted from January 1, 2011, to November 30, 2020.It adhered to the Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines.

Fig 1 .
Fig 1. Flow diagram of the study population

Table 1
Baseline sample characteristics before and after inverse probability weighting b Abbreviations: SCD=sudden cardiac death; CCI=Charlson comorbidity index; DM=diabetes mellitus; IPW=inverse probability weighting; NT $=new Taiwan dollar; SD=standard deviate; TdP=torsade de pointes; SMD=standardized mean difference.a The cases and controls were frequency-matched at a ratio of 1:4 by age, sex, and the new diagnosis year.b The propensity score used for all characteristics, including age, sex, diagnosed year, CCI score, comorbidity and co-medications.c

Table 2
Risk of sudden cardiac death among antipsychotic formulations in patients with newly diagnosed schizophrenia.

Table 3
Sensitivity risk of sudden cardiac death after treatment within the 30, 90, and 180 days.SCD=sudden cardiac death; OAP=oral antipsychotic; (Ever-used) LAIs=LAI monotherapy or LAI and OAP combination; LAI=long-acting injectable antipsychotic; Non-antipsychotic user=neither OAPs nor LAIs within 60 days; IPW=inverse probability weighting; OR=odd ratio; CI=confidence interval.The aOR was derived from the logistic regression model adjusted for age, sex, geographic region, urban level, month income, CCI score, comorbidity, co-medications, and number of hospitalizations. a