Enhancing Suicide Risk Prediction With Polygenic Scores in Psychiatric Emergency Settings: Prospective Study

Area of Science:

• Psychiatric genetics
• Clinical prediction modeling
• Genomic risk assessment

Background:

• Polygenic risk scores (PRSs) show promise for predicting psychiatric outcomes.
• The clinical utility of integrating PRSs into existing prediction models remains unclear.
• Electronic health records (EHRs) and surveys are commonly used for risk assessment.

Purpose of the Study:

• To evaluate the clinical utility of incorporating PRSs into a suicide risk prediction model.
• The model was trained on EHR data and patient surveys for emergency department patients.
• Assessed if genomic information enhances prediction beyond clinical data alone.

Main Methods:

• Recruited 333 adult patients of European ancestry from a psychiatric emergency department.
• Generated multiple neuropsychiatric PRSs (depression, bipolar disorder, schizophrenia, suicide attempt, externalizing traits).
• Added PRSs to a validated suicide prediction model using EHRs and surveys; evaluated using AUC and precision-recall curves.

Main Results:

• 8.4% of patients had a suicide attempt within 6 months.
• Adding schizophrenia PRS or all PRSs numerically improved discrimination (AUC 0.86) compared to the baseline model (AUC 0.84).
• The observed improvement in model performance was not statistically significant.

Conclusions:

• Incorporating genomic information via PRSs did not significantly improve suicide attempt risk stratification in this cohort.
• Larger, more diverse studies are needed to validate the role of psychiatric PRSs in clinical prediction models.
• The current study suggests limited added value of PRSs for suicide risk prediction in this specific population.