Background: There have been few attempts formally to validate screening measures for postnatal common mental disorder in low income country settings. We have investigated the comparative validity of three different screening approaches in a community-based study in Kintampo, Ghana.
Method: 160 women aged 15-45 years, and 5-11 weeks postpartum were first screened using the Self-Report Questionnaire (SRQ-20), with oversampling of higher scorers. The other test assessments were the Edinburgh Postnatal Depression Scale (EPDS) and the Patient Health Questionnaire (PHQ-9). Criterion validity was measured against the Comprehensive Psychopathological Rating Scale (CPRS), and concurrent validity against the WHO Disability Assessment Schedule. A sub-sample (n=40) was re-interviewed 2 weeks later for test-retest reliability.
Results: Internal consistency (Cronbach's Alpha) was equivalent across all three test scales; EPDS (0.79), SRQ-20 (0.78) and PHQ-9 (0.79). Test-retest reliability was better for PHQ-9 (ICC 0.75) than for the EPDS (0.51). For criterion validity the PHQ-9 (AUROC 0.90 (0.81-0.98)), was superior to the SRQ-20 (0.74 (0.62-0.86)) and the EPDS ((0.84 (0.76-0.92). Youden's Index was also superior for PHQ-9. Item analysis revealed that a mixture of somatic and cognitive symptoms best discriminated between cases and non-cases for all three scales.
Limitations: Inability to ascertain inter-rater reliability, order effects and possible loss of technical equivalence due to item modifications.
Conclusions: The evidence for the validity, reliability, and superiority of the PHQ-9 over other screening assessments has been extended. The PHQ-9 is short, easy to administer and acceptable to a largely illiterate population of Ghanaian women, 5 to 11 weeks post partum.