Per PI, this study is not a clinical trial and was inadvertently entered in the system

The objective of this retrospective observational study is to compare commonly prescribed bipolar disorder medications for their impact on: (1) hospitalization; (2) suicide attempts and self-harm; and (3) risk of drug-induced adverse effects such as kidney disease and diabetes mellitus. In addition, the investigators will examine heterogeneity of treatment effect by co-morbidity within pediatric, adult, and elderly sub-populations. Patient focus groups are convened to elicit additional questions and provide feedback on results.

Funded by PCORI, the objective of this retrospective observational study is to perform several safety and effectiveness comparisons on commonly prescribed bipolar disorder medications, engaging patient focus groups in generating additional questions and interpreting results.

The study will be a retrospective cohort study conducted with administrative claims data from the Truven MarketScan Commerical Claims and Encounters and Medicare database from 2010-2016.

The database contains approximately 140 million patients within the US population in every state and nearly every county in the nation, across all ages, ethnicities and socioeconomic categories, including privately insured, and Medicare patients. The study will focus on approximately one million patients with two or more diagnoses of bipolar disorder in the claims records according to ICD-9 and/or ICD-10 coding.

The treatments that will be compared are lithium carbonate; first generation antipsychotics: haloperidol and perphenazine; second generation antipsychotics: clozapine, risperidone, olanzapine, aripiprazole, quetiapine, ziprasidone, asenapine, lurasidone, and paliperidone; mood stabilizing anticonvulsants: valproate, lamotrigine, carbamazepine, and oxcarbazepine; antidepressants: mirtazapine, bupropion, desvenlafaxine, duloxetine, venlafaxine, citalopram, escitalopram, fluoxetine, fluvoxamine, paroxetine, sertraline, vilazodone, and doxepin.

The investigators will perform cross-sectional and survival based analysis using regression, propensity scoring, and local control to perform bias-corrected comparisons of the above treatments for for their impact on: (1) hospitalization; (2) suicide attempts and self-harm; and (3) risk of drug-induced adverse effects such as kidney disease and diabetes mellitus. In addition, the investigators will examine heterogeneity of treatment effect by co-morbidity within pediatric, adult, and elderly sub-populations.

Inclusion Criteria:

• Two or more instances of bipolar disorder diagnoses within administrative claims records

Exclusion Criteria:

• Patients with less than 1 year of history in the database

The study population consists of individuals with bipolar disorder enrolled with private insurers or with Medicare across the United States, as captured in the Truven Health Analytics MarketScan databases. The time period of data coverage is 2003-2016.

This retrospective observational study makes use of MarketScan individual level patient data from Truven Health Analytics. The license terms for data access prohibit public dissemination of individual level patient data. The investigators will, however, make openly available the queries and analytic procedures required to reproduce the study.