An Emulation of Randomized Trials of Administrating Antipsychotics in PTSD Patients for Outcomes of Suicide-Related Events

Post-traumatic stress disorder (PTSD) is a prevalent mental disorder marked by psychological and behavioral changes. Currently, there is no consensus of preferred antipsychotics to be used for the treatment of PTSD. We aim to discover whether certain antipsychotics have decreased suicide risk in the PTSD population, as these patients may be at higher risk. A total of 38,807 patients were identified with a diagnosis of PTSD through the ICD9 or ICD10 codes from January 2004 to October 2019. An emulation of randomized clinical trials was conducted to compare the outcomes of suicide-related events (SREs) among PTSD patients who ever used one of eight individual antipsychotics after the diagnosis of PTSD. Exclusion criteria included patients with a history of SREs and a previous history of antipsychotic use within one year before enrollment. Eligible individuals were assigned to a treatment group according to the antipsychotic initiated and followed until stopping current treatment, switching to another same class of drugs, death, or loss to follow up. The primary outcome was to identify the frequency of SREs associated with each antipsychotic. SREs were defined as ideation, attempts, and death by suicide. Pooled logistic regression methods with the Firth option were conducted to compare two drugs for their outcomes using SAS version 9.4 (SAS Institute, Cary, NC, USA). The results were adjusted for baseline characteristics and post-baseline, time-varying confounders. A total of 5294 patients were eligible for enrollment with an average follow up of 7.86 months. A total of 157 SREs were recorded throughout this study. Lurasidone showed a statistically significant decrease in SREs when compared head to head to almost all the other antipsychotics: aripiprazole, haloperidol, olanzapine, quetiapine, risperidone, and ziprasidone (p < 0.0001 and false discovery rate-adjusted p value < 0.0004). In addition, olanzapine was associated with higher SREs than quetiapine and risperidone, and ziprasidone was associated with higher SREs than risperidone. The results of this study suggest that certain antipsychotics may put individuals within the PTSD population at an increased risk of SREs, and that careful consideration may need to be taken when prescribed.

Abstract: Post-traumatic stress disorder (PTSD) is a prevalent mental disorder marked by psychological and behavioral changes. Currently, there is no consensus of preferred antipsychotics to be used for the treatment of PTSD. We aim to discover whether certain antipsychotics have decreased suicide risk in the PTSD population, as these patients may be at higher risk. A total of 38,807 patients were identified with a diagnosis of PTSD through the ICD9 or ICD10 codes from January 2004 to October 2019. An emulation of randomized clinical trials was conducted to compare the outcomes of suicide-related events (SREs) among PTSD patients who ever used one of eight individual antipsychotics after the diagnosis of PTSD. Exclusion criteria included patients with a history of SREs and a previous history of antipsychotic use within one year before enrollment. Eligible individuals were assigned to a treatment group according to the antipsychotic initiated and followed until stopping current treatment, switching to another same class of drugs, death, or loss to follow up. The primary outcome was to identify the frequency of SREs associated with each antipsychotic. SREs were defined as ideation, attempts, and death by suicide. Pooled logistic regression methods with the Firth option were conducted to compare two drugs for their outcomes using SAS version 9.4 (SAS Institute, Cary, NC, USA). The results were adjusted for baseline characteristics and post-baseline, time-varying confounders. A total of 5294 patients were eligible for enrollment with an average follow up of 7.86 months. A total of 157 SREs were recorded throughout this study. Lurasidone showed a statistically significant decrease in SREs when compared head to head to almost all the other antipsychotics: aripiprazole, haloperidol, olanzapine, quetiapine, risperidone, and ziprasidone (p < 0.0001 and false discovery rate-adjusted p value < 0.0004). In addition, olanzapine was associated with higher SREs than quetiapine and risperidone, and ziprasidone was associated with higher SREs than risperidone. The results of this study suggest that certain antipsychotics may put individuals within the PTSD population at an increased risk of SREs, and that careful consideration may need to be taken when prescribed.

Relationship between Antipsychotics and PTSD
Antipsychotics, in the context of treatment for PTSD, have typically been used to augment the effects of first-line agents such as serotonin receptor inhibitors (SRIs) or as second-line treatment [20]. Small randomized controlled trials have been conducted to compare the efficacy of monotherapy SGAs, such as quetiapine and olanzapine, in patients with PTSD [21,22]. Results were limited but show promising efficacy for the treatment of PTSD with SGAs.
Currently, there is no consensus of preferred antipsychotics to be used for the treatment of PTSD. The goal of this observational study is to identify specific antipsychotics that may be associated with a decreased incidence of SREs as defined above.

Emulating Randomized Controlled Trial
Creating guidance for clinical guidelines requires sound evidence on the effects of their recommendations on relevant outcomes. The golden standard of investigation relies heavily on randomized controlled trials. However, randomized controlled trials are not always feasible due to money, time, or ethical constraints. Observational data are often used as an alternative to randomized controlled trials.
Studies using observational data are subject to selection and immortal-time bias, which can be controlled in randomized controlled trials [23]. In this study, we use emulation methods to control these two biases. Emulation attempts to eliminate selection bias by specifying a target trial and creating an appropriate protocol [24]. This allows for the comparison of individuals newly initiated on the medication against those who were never initiated. Immortal-time bias can be minimized by analyzing the data so that time of eligibility for treatment and time when treatment is initiated is the same [25].

Data Source
We accessed data from January 2004 to October 2019 from the Neptune system at University of Pittsburgh which manages use of patient electronic medical records from the UPMC health system for research purposes (rio.pitt.edu/services accessed on 3 March 2021). The database includes demographic information, diagnoses, encounters, medication prescriptions, prescription fill history, and laboratory tests.

Inclusion/Exclusion Criteria and Endpoints/Follow Up
The baseline eligibility criteria include initiated an antipsychotic after the diagnosis of PTSD (qualifying event), no history of suicidal-related events prior to the enrollment, no prior use of antipsychotics within one year, and at least one year of history in the electronic medical record. If eligibility criteria were met, patients were followed to the onset of the first suicide-related event (primary outcome) or until loss to follow up (stopping use of antipsychotic, switching to another antipsychotic, patients data being no longer accessible or reaching the end time of this study).

Target Trials
In order to be included in the trial, eligible PTSD patients must have no experience of SREs and no usage of any antipsychotic during one year prior to the enrollment. Patients were randomly assigned to one of the compared antipsychotic arms in the target trial. Antipsychotics of interest include aripiprazole, haloperidol, lurasidone, olanzapine, quetiapine, risperidone, perphenazine, and ziprasidone. Patients could take any other medications for the treatment of their comorbidities. The trial will be stopped if a patient stops using the drug assigned or switches or starts using another antipsychotic, loses follow up, or experiences SREs. The outcome evaluated is the onset of SREs.

Emulating Target Trials
We attempted to emulate randomized controlled trials similarly to the work of Danaei and colleagues [26]. Confounding variables had to be measured at least once during the study duration. These variables can be accessed below in Table 1 of methods section and categories are based on ICD9 and ICD10 codes described in Appendix A. In addition to the baseline eligibility criteria that is discussed above, participants needed one or more years of continuous recording in the UPMC medical records and at least one medical visit within one year of initiation of trial.
Monthly trials were collected from the UPMC EMR database from January 2004 to October 2019 (190 months of interval). Patients could be included in this study multiple times if a one-year washout period of antipsychotic was fulfilled and all inclusion criteria were met. Those eligible were placed in a specific target trial arm based on the antipsychotic used. The target trials of antipsychotics were then compared with the primary outcome being SREs experienced. Study duration was stopped if medication of interest was discontinued or if the patient's EMR data were no longer available (loss to follow up or death). In other words, patients were right censored if they had no SREs at the end of the study period (i.e., administrative censoring), if a person failed to return for a study visit (i.e., lost to follow up), if they stopped using the antipsychotic of interest (no records of use within 3 months), or if they switched to another antipsychotic.

Per-Protocol Analysis
Per-protocol analysis was utilized which requires that all the patients fully completed the given protocol regime. The effect of the two drugs was compared on the outcomes of two cohorts who completed the treatment originally allocated. This analysis may give rise to bias by baseline confounders and post-baseline, time-varying confounders. An approach suggested by Danaei and colleagues used a pooled logistic regression model in order to estimate the effect of treatment [25]. They used inverse probability weighting to create a population where treatment is independent of prognostic factors history. The baseline information included 12 categories of mental disorders (Appendix B) [27], age, gender, number of emergency department visits within one year prior to the enrollment and the summary can be found in the Table 1. To test the effects of concomitant medications, we performed an additional analysis by adjusting the most frequently used drugs of the central nervous system in drug pair(s) of interest.
In our emulation, a pooled logistic regression model was fitted to adjust the censoring effects. Switching from one of the paired antipsychotics to the other one was not considered since we believe this situation can be represented by the censor model.
In addition, we used robust variances to calculate conservative 95% confidence intervals and truncated the inverse-probability weights to their 99th percentile. Those are options provided by the code from www.hsph.harvard.edu/causal/software (accessed on 27 November 2020). We also applied the Firth option in the logistic regression to accommodate the presence of rare events or complete separation. Datasets were prepared by Python [28] and the final analyses were carried out using SAS 9.4 (SAS Institute Inc., Cary, NC, USA) [29]. False discovery rate (FDR) q value is controlled to 0.05 to avoid the inflation of type I error cause by multiple hypothesis test. The calculation of FDR is conduct by R [30] 4.0.2 base package function "p.adjust".

Results
Of the 38,807 patients diagnosed with PTSD that were identified, only 5294 patients (4901 unique patients) were initiated on an antipsychotic after the diagnosis of PTSD, had no history of suicidal-related events prior to the enrollment, and no prior use of antipsychotics within one year, with at least one year of history in the electronic medical record. Eligible individuals were enrolled in this study and were assigned to a treatment group and followed until stopping current treatment, switching to another same class of drugs, death or loss to follow up. Quetiapine was the most commonly assigned antipsychotic, consisting Of the patients enrolled in this study, 5137 patients were right censored. The average follow up of the antipsychotics was 7.86 months, ranging from 4.38 to 9.76 months. Patients taking quetiapine (1429 patients) or aripiprazole (982 patients) experienced the most amount of SREs, as 38 events were recorded for both groups. Zero patients taking lurasidone experienced an SRE ( Figure 2).

Figure 2.
Enrollment process for emulation. Patients were right censored if they had no SREs at the end of the study period (i.e., administrative censoring), if a person failed to return for a study visit (i.e., lost to follow up), if they stopped using the antipsychotic of interest (no records of use within 3 months), or if they switched to another antipsychotic. SRE: suiciderelated events.  Of the patients enrolled in this study, 5137 patients were right censored. The average follow up of the antipsychotics was 7.86 months, ranging from 4.38 to 9.76 months. Patients taking quetiapine (1429 patients) or aripiprazole (982 patients) experienced the most amount of SREs, as 38 events were recorded for both groups. Zero patients taking lurasidone experienced an SRE ( Figure 2).
Results of the emulations are shown in Table 2. Head-to-head comparisons are conducted between every pair of antipsychotics to compare their effectiveness in managing SREs. As shown in Table 2, if only adjusting for comorbidities (and age, gender, ED visits) lurasidone showed a significantly better effect compared to all other antipsychotics while ziprasidone and olanzapine were associated with higher SREs in patients.
To assess the effects of using concomitant medications on the SREs, we also adjusted for the most used medications. Those concomitant drugs are central nervous system agents and had been used at the baseline time in more than 5% of users of an antipsychotic drug of interest. To reduce number of variables in the models, those comorbidities with less than 5% patients in the paired antipsychotics were not included. After further adjusting those drugs, the hazard ratio of lurasidone against other antipsychotics are still significantly better with p values less than 0.0001 and adjusted p values less than 0.04 (except for perphenazine vs. lurasidone, which has no significant difference). We also noticed that aripiprazole vs. lurasidone has HR of 43.25 and haloperidol vs. lurasidone has HR of 52.20. These HRs have large confidence intervals (CI). This might indicate that the models become unstable when more covariates were added. Detailed information on all the concomitant medications can be found in Appendix C.
Of the patients enrolled in this study, 5137 patients were right censored. The average follow up of the antipsychotics was 7.86 months, ranging from 4.38 to 9.76 months. Patients taking quetiapine (1429 patients) or aripiprazole (982 patients) experienced the most amount of SREs, as 38 events were recorded for both groups. Zero patients taking lurasidone experienced an SRE ( Figure 2).    Figure 3 showed the standardized survival curves for lurasidone vs. olanzapine, risperidone vs. ziprasidone and olanzapine vs. risperidone. Please note, the curves were the results of adjusting covariates during the comparison of two paired drugs, and as such these curves do not necessarily reflect the real SREs listed above. The survival curves were generated according to the work of Danaei and colleagues [26]. We estimated survival curves for each arm by fitting a pooled logistic model and included product ("interaction") terms to allow the hazard ratio to change over time.

Figure 3.
Examples of survival curves for three paired drugs on the outcome of SREs. X axis is months of follow up and y axis is the survival rate. These curves were generated by adjusting the baseline and post-baseline variables and may not reflect the real data. SREs: suicide-related events.

Discussion and Conclusions
The primary objective of this observational study was to identify specific antipsychotics that may be associated with a decreased incidence of suicide-related events. In this study, we emulated a randomized controlled trial using the Neptune system at University of Pittsburgh with UPMC health system electronic medical records from January 2009 to October 2019. We utilized baseline characteristic data to assess suicide risk in patients diagnosed with PTSD. Patients were assigned one of the compared antipsychotics and the suicidal risk was statistically analyzed.
Only a handful of studies have been conducted on antipsychotics for the treatment of PTSD, and even fewer on the medications' effect on SREs [33]. As discussed, the results of the limited studies do show promising efficacy for the treatment of PTSD with SGAs [21,22]. Lurasidone, one of the less frequently prescribed antipsychotics, showed a statistically significant decrease in SREs compared to all the other antipsychotics: aripiprazole, haloperidol, olanzapine, quetiapine, risperidone, and ziprasidone (except perphenazine) ( Table 1). In addition, olanzapine was associated with higher SREs than quetiapine and risperidone, and ziprasidone was associated with higher SREs than risperidone. The results of this study indicate that commonly used antipsychotics such as olanzapine may put individuals within the PTSD population at an increased risk of SREs. The findings of this study suggest that special considerations may need to be put in place whenever prescribing or recommending these medications to PTSD patients.
Antipsychotics have a variety of indications both FDA and non-FDA approved [34]. While the majority of medications in the class are indicated for schizophrenia and bipolar disorder, some other uses include Tourette's disorder, nausea and vomiting, and chronic pain. Lurasidone is FDA approved for the treatment of schizophrenia as well as depression associated with bipolar disorder, or manic-depressive disorder, as adjunct therapy [35]. The findings of our study suggest that this antipsychotic may decrease the risk of SREs, and this information may be useful when recommending patient-specific pharmacotherapy to patients with PTSD.
Ziprasidone, another less frequently prescribed antipsychotic, showed statistically significant higher SREs than risperidone. The atypical antipsychotic is FDA approved for the treatment of both bipolar disorder and schizophrenia. Hamner et al. conducted a randomized, 2-phase, placebo-controlled study of ziprasidone efficacy in refractory PTSD patients [36]. Patients refractory to SSRI treatment in phase 1 were assigned either ziprasidone or placebo. No statistically significant differences were discovered when ziprasidone was given to PTSD patients refractory to initial SSRI treatment, compared to placebo. As alternatives for PTSD treatment arise, pilot studies on antipsychotics such as this one are

Discussion and Conclusions
The primary objective of this observational study was to identify specific antipsychotics that may be associated with a decreased incidence of suicide-related events. In this study, we emulated a randomized controlled trial using the Neptune system at University of Pittsburgh with UPMC health system electronic medical records from January 2009 to October 2019. We utilized baseline characteristic data to assess suicide risk in patients diagnosed with PTSD. Patients were assigned one of the compared antipsychotics and the suicidal risk was statistically analyzed.
Only a handful of studies have been conducted on antipsychotics for the treatment of PTSD, and even fewer on the medications' effect on SREs [33]. As discussed, the results of the limited studies do show promising efficacy for the treatment of PTSD with SGAs [21,22]. Lurasidone, one of the less frequently prescribed antipsychotics, showed a statistically significant decrease in SREs compared to all the other antipsychotics: aripiprazole, haloperidol, olanzapine, quetiapine, risperidone, and ziprasidone (except perphenazine) ( Table 1). In addition, olanzapine was associated with higher SREs than quetiapine and risperidone, and ziprasidone was associated with higher SREs than risperidone. The results of this study indicate that commonly used antipsychotics such as olanzapine may put individuals within the PTSD population at an increased risk of SREs. The findings of this study suggest that special considerations may need to be put in place whenever prescribing or recommending these medications to PTSD patients.
Antipsychotics have a variety of indications both FDA and non-FDA approved [34]. While the majority of medications in the class are indicated for schizophrenia and bipolar disorder, some other uses include Tourette's disorder, nausea and vomiting, and chronic pain. Lurasidone is FDA approved for the treatment of schizophrenia as well as depression associated with bipolar disorder, or manic-depressive disorder, as adjunct therapy [35]. The findings of our study suggest that this antipsychotic may decrease the risk of SREs, and this information may be useful when recommending patient-specific pharmacotherapy to patients with PTSD.
Ziprasidone, another less frequently prescribed antipsychotic, showed statistically significant higher SREs than risperidone. The atypical antipsychotic is FDA approved for the treatment of both bipolar disorder and schizophrenia. Hamner et al. conducted a randomized, 2-phase, placebo-controlled study of ziprasidone efficacy in refractory PTSD patients [36]. Patients refractory to SSRI treatment in phase 1 were assigned either ziprasidone or placebo. No statistically significant differences were discovered when ziprasidone was given to PTSD patients refractory to initial SSRI treatment, compared to placebo. As alternatives for PTSD treatment arise, pilot studies on antipsychotics such as this one are crucial. The efficacy and safety of ziprasidone in both the study by Hammer and company and our study are less than ideal for the target population. Prescribers considering alternate therapies for PTSD patients may want to take careful consideration of ziprasidone until further, larger, studies can be completed.
Many studies and publications have covered antipsychotic drugs and the impact on SREs in schizophrenic patients, but the same relationship is scarcely trialed in the PTSD population [37]. Katz et al. conducted a study on suicidal risk in patients with both PTSD and bipolar disorder [38]. The multiple regression analysis identified that comorbid PTSD is a significant predictor and increases risk of suicidal ideation in bipolar patients. This study confirms the demand for more pharmacologic research for this patient population. With many antipsychotics indicated for the treatment of bipolar disorder, patients with comorbid PTSD may benefit from a specific antipsychotic that can lower the risk of SREs. If a physician is choosing an antipsychotic for a patient with PTSD, bipolar disorder, and a risk/history of SREs, recommending lurasidone may be a viable option.
One of the other drawbacks of this study is the validity and usefulness of the lurasidone results. Though the medication displayed statistically significant fewer SREs than several more frequently prescribed antipsychotics, lurasidone had not been prescribed to PTSD patients since 2012 in the UPMC database systems. On the contrary, all the other antipsychotics included in this study showed an increase in enrollment since 2012. Though popular in the years of 2008 to 2012, lurasidone has had a sharp decline in use since that time period. One hypothesis to this observation points to modification in insurance company authorization. Lurasidone is an expensive medication with an estimated monthly cost of $1500. This high cost may have kept insurances wary of covering lurasidone as a first-line treatment before trial of several other less expensive antipsychotics. Insurance companies likely required the failure of two generically available antipsychotics before agreeing to cover lurasidone. Furthermore, most insurance companies would have likely required the diagnosis of schizophrenia or bipolar disorder. We further investigated the number of patients who switched their assignments and the number of ED visits at the first and the last month of follow ups. Our assumption is that if a drug is right censored because of its side effects or lack of therapeutic effects, the percentage of switching and/or number of ED visits would be higher than other drugs. Lurasidone users had comparable number of months of follow up, percentage of switching and average number of ED visits with users of other antipsychotics (Appendix C). We believe that the right-censoring of lurasidone is unlikely because of greater adverse effects or lack of efficacy.
Our study also has some limitations. Firstly, as this is not a randomized control study, we cannot rule out the possibility of uncontrolled biases. For example, socioeconomic status is an important factor for predicting SREs, but due to the limitation of our EMR dataset, such information is difficult to access. However, medication usage information was controlled which might mitigate the influence of some biases. In our recent publication, we predicted patients at increased risk for SREs with high accuracy through utilizing the demographic, comorbidities, and medication use information from EMR, without socioeconomic information [13]. Secondly, we defined SREs as ideation, attempts, and death by suicide. Because those three behaviors have very different constructs, it would be more suitable to use separate models [39]. However, in our study population of PTSD patients, which is a subpopulation of patients with high risk for SREs, the number of events is still low. To reach the statistical significance level, a larger number of patients is needed which is not feasible at this current stage. On the other hand, our study can be interpreted as the influence of antipsychotics on the risk shared among the ideations, attempts, and deaths by suicide. Thirdly, a great limitation of this work is the lack of any psychometric instrument concerning the severity of the symptoms of PTSD and lack of data on the motivation that led to the prescription of antipsychotics. Again, this is also related to the limitation of our EMR data and such information can be found in an unstructured data form such as medical notes which is a challenge to extract.
Antipsychotics, in the context of treatment for PTSD, have typically been used as second-line agents as monotherapy or to augment the effects of first-line agents such as serotonin receptor inhibitors (SRIs). For patients who decline the use of antidepressants or psychotherapy for PTSD treatment, antipsychotics have emerged as a viable option. In our study, we controlled the use of antidepressants as two antidepressants paroxetine and sertraline are the first-line medications for PTSD. Our data showed that only a small portion of those patients were using those two medications at baseline. For example, the percentages of patients with coadministration of sertraline were in the range of 12.4% to 18.2% (Appendix C), while paroxetine usage was in the range of 1.0% to 5.2%. Limited studies have confirmed the effectiveness and safety of antipsychotics in the PTSD population, and there are currently no studies with lurasidone. With the current lack of knowledge and research of antipsychotics for PTSD patients, specifically lurasidone, we hope our study provides a critical gateway into the relationship between certain antipsychotics and suicide risk.

Data Availability Statement:
The data used in this study was from UPMC under a data use agreement. The authors are not allowed to distribute the data to any third party, but researchers may contact UPMC for data access.  Disturbance of emotions specific to childhood and adolescence