Psychiatric Diagnostic Screening Questionnaire (PDSQ): Psychometric Properties and Item-Level Empirical Correlates

To investigate scale-level and especially item-level correlates of the Psychiatric Diagnostic Screening Questionnaire (PDSQ) with an emphasis on its clinical usefulness in primary care settings. In addition, gender-based correlate differences were explored to test whether gender-specific conclusions were warranted. Overall, the current empirical correlates sought to improve the PDSQ's precision in clinical interpretations.

The current dataset was composed of 234 primary care patients, systematically evaluated via the Structured Clinical Interview for DSM-5 Disorders-Research Version (SCID-5-RV) for 10 common mental disorders. Rather than using the typical skip-out logic, the 10 modules were administered in their entirety, providing item-level correlates for the PDSQ as the index measure.

When adequately represented (n ≥ 10), PDSQ subscales produced strong scale homogeneities (M alpha = 0.85). At the item level, they evidenced strong convergent validities (M r = 0.64), clearly more than doubled their discriminant validities (M r = 0.25). Adopting a stringent criterion (r ≥ 0.35), 273 PDSQ empirical correlates were established for common diagnoses. Regarding gender, seven disorders produced similar scale correlates, whereas three were divergent.

As the third independent investigation, PDSQ item-level correlates have been empirically validated in both psychiatric and primary care settings. The PDSQ, in its expanded role, provided valuable clinical data regarding DSM-5 symptoms and clinical characteristics.

Consistent with the first objective, the PDSQ subscales demonstrated excellent internal consistency and construct validity. For specific symptoms and clinical characteristics, the PDSQ is distinguished from most self-report psychological measures regarding the precision of its clinical findings via item-level correlates. In the context of the replication crisis, rigorous criteria were applied to bivariate correlates, ranging from moderate (rs ≥ 0.35) to very large (rs ≥ 0.60) for each PDSQ subscale.