Computer adaptive testing strategies for the Edinburgh Postnatal Depression Scale (EPDS).

Area of Science:

• Psychiatry
• Psychometrics
• Public Health

Background:

• Perinatal Mood and Anxiety Disorders (PMADs) affect parent, child, and family well-being, with disparities in diagnosis and treatment among minority groups.
• Existing screening methods for PMADs can be lengthy, potentially hindering widespread adoption and equitable access.
• Computer Adaptive Testing (CAT) offers a method to improve screening efficiency by reducing test length.

Purpose of the Study:

• To evaluate if CAT-based short-form versions of the Edinburgh Postnatal Depression Scale (EPDS) maintain diagnostic accuracy.
• To determine if these CAT methods exacerbate existing racial disparities in PMAD screening outcomes.

Main Methods:

• Real data simulation was used to assess three CAT-based short-form EPDS versions derived from item response theory models.
• Diagnostic precision of the short forms was evaluated.
• Potential racial disparities in false negative rates were examined compared to the full-length EPDS.

Main Results:

• CAT-based EPDS short forms showed high correlation (r=0.96-0.97) with the full-length EPDS, exceeding previous CAT-MDD correlations (r=0.82).
• False negative rates for CAT-implied diagnoses did not significantly differ across racial groups, indicating no racial bias.
• CAT application to EPDS maintains diagnostic precision.

Conclusions:

• CAT-based EPDS versions are a promising tool for efficient PMAD screening.
• These tools maintain diagnostic precision and do not worsen racial disparities.
• Reduced evaluation time can facilitate wider, more equitable screening for earlier PMAD diagnosis and treatment.