Exploring the predictive validity of personality disorder criteria.

Area of Science:

• Psychiatry
• Psychology
• Clinical Psychology

Background:

• Personality disorders (PDs) are complex mental health conditions.
• Understanding the predictive validity of PD indicators is crucial for clinical assessment and treatment planning.

Purpose of the Study:

• To evaluate the predictive validity of personality disorder indicators at varying levels of aggregation.
• To compare the predictive accuracy of models using individual PD criteria versus aggregated scales.

Main Methods:

• Utilized interview data on 78 DSM-IV PD criteria across three large samples (N=651, N=552, N=1,277).
• Employed regularized linear regression and regression forest algorithms in a repeated cross-validation design.
• Assessed out-of-sample performance and model predictions based on aggregated and item-level data.

Main Results:

• No significant difference in predictive performance was found between models trained on item-level data and those trained on scale-level data.
• Little predictive variance was lost when aggregating individual DSM-IV PD criteria into dimensional PD scales.
• Model performance remained consistent across different samples and aggregation levels.

Conclusions:

• Dimensional personality disorder scales retain significant predictive validity.
• Aggregation of PD criteria to scales is a viable approach for predictive modeling in clinical psychology.
• Findings support the use of dimensional scales in assessing and predicting outcomes related to personality pathology.