Using computerised adaptive tests to screen for perinatal depression in underserved women of colour.

Background: Compared with traditional screening questionnaires, computerised adaptive tests for severity of depression (CAT-DI) and computerised adaptive diagnostic modules for depression (CAD-MDD) show improved precision in screening for major depressive disorder. CAT measures have been tailored to perinatal women but have not been studied in low-income women of colour despite high rates of perinatal depression (PND).

Objective: This study aimed to examine the concordance between CAT and traditional measures of depression in a sample of primarily low-income black and Latina women.

Methods: In total, 373 women (49% black; 29% Latina) completed the Patient Health Questionnaire-9 (PHQ-9), CAD-MDD and CAT-DI at 845 visits across pregnancy and postpartum. We examined the concordance between continuous CAT-DI and PHQ-9 scores and between binary measures of PND diagnosis on CAD-MDD and the PHQ-9 (cut-off score >10). We examined cases with a positive PND diagnosis on the CAD-MDD but not on the PHQ-9 ('missed' cases) to determine whether clinic notes were consistent with CAD-MDD results.

Findings: CAT-DI and PHQ-9 scores were significantly associated (concordance correlation coefficient=0.67; 95% CI 0.58 to 0.74). CAD-MDD detected 5% more case of PND compared with PHQ-9 (p<0.001). The average per-visit rate of PND was 14.4% (14.5% in blacks, 14.9% in Latinas) on the CAD-MDD, and 9.5% (9.8% in blacks, 8.8% in Latinas) on the PHQ-9. Clinical notes were available on 60% of 'missed' cases and validated CAD-MDD PND diagnosis in 89% of cases.

Conclusions: CAD-MDD detected 5% more cases of PND in women of colour compared with traditional tests, and the majority of these cases were verified by clinician notes.

Clinical implications: Use of CAT in routine clinic care may address health disparities in PND screening.