Related Experiment Video
Updated: Nov 14, 2025

Author Spotlight: Therapeutic Benefit of Closed-Loop Deep Brain Stimulation in Depression Treatment
Published on: July 7, 2023
Using Machine Learning Imputed Outcomes to Assess Drug-Dependent Risk of Self-Harm in Patients with Bipolar Disorder:
Anastasiya Nestsiarovich1, Praveen Kumar2, Nicolas Raymond Lauve2
1Center for Global Health, Department of Internal Medicine, University of New Mexico Health Sciences Center, Albuquerque, NM, United States.
This study found that certain antidepressants, mood-stabilizing anticonvulsants (MSAs), and psychotherapy effectively prevent self-harm in bipolar disorder (BD). Some drug combinations increased self-harm risk compared to lithium.
Area of Science:
- Psychiatry
- Pharmacology
- Health Informatics
Background:
- Observational studies on self-harm prevention in bipolar disorder (BD) are hindered by incomplete suicidality coding in administrative claims data.
- Assessing pharmacotherapy and psychotherapy effectiveness for self-harm prevention in BD is challenging due to missing outcome data.
- Conflicting evidence exists regarding drug-dependent self-harm risk and the preventive effects of various treatment strategies in BD.
Purpose of the Study:
- To compare the risk of self-harm across commonly used BD pharmacotherapies and psychotherapy in a large insured population.
- To address underrecording of self-harm events in US electronic health records through imputation methods.
- To provide evidence-based insights into effective self-harm prevention strategies for bipolar disorder.
Main Methods:
- Utilized the IBM MarketScan administrative claims database to analyze self-harm risk in BD patients across 65 drug regimens and drug-free periods.
- Employed machine learning for imputing probable but uncoded self-harm events, with sensitivity analyses on probability thresholds.
- Applied Cox regression models to compare risks associated with various pharmacotherapies (lithium, MSAs, SGAs, FGAs, antidepressants) and psychotherapy.
Main Results:
- Among 529,359 patients, 1.66% experienced self-harm events. Adolescence was associated with higher self-harm risk.
- Six regimens showed increased self-harm risk versus lithium, including certain antidepressant + second-generation antipsychotic (SGA) combinations.
- Nine regimens demonstrated lower self-harm risk, including lamotrigine, valproate, risperidone, aripiprazole, and specific antidepressants (SNRI, SSRI, bupropion). Psychotherapy alone significantly reduced self-harm risk compared to no treatment.
Conclusions:
- Findings support the efficacy of antidepressants, mood-stabilizing anticonvulsants (MSAs), and psychotherapy in preventing self-harm among individuals with bipolar disorder (BD).
- Certain medication combinations are associated with a higher risk of self-harm compared to lithium, highlighting the importance of regimen selection.
- Machine learning imputation of self-harm events proved effective in overcoming data limitations for observational studies in BD.
Related Concept Videos
Mania and Antimanic Drugs: Overview
Types of Biopharmaceutical Studies: Controlled and Non-Controlled Approaches
Non-controlled studies, commonly employed for initial exploration, lack a control group, rendering them susceptible to biases and external influences. In contrast,...
Bipolar Disorder

