Table 2.
Comparison of major depression classification odds across diagnostic interviews
| Diagnostic interview comparison | Screening tool | Model without interactiona | Model with interactionb | |
|---|---|---|---|---|
| Adjusted odds ratio aOR (95% CI) | Adjusted odds ratio aOR (95% CI) | Adjusted odds ratio aOR for interaction (95% CI) | ||
| CIDI vs. SCID | PHQ-9 | 0.81 (0.50, 1.33) | 1.15 (0.69, 1.92) | 0.73 (0.66, 0.81) |
| EPDSc | 0.34 (0.09, 1.34) | 0.66 (0.15, 2.82) | 0.50 (0.41, 0.61) | |
| HADS-Dc | 1.09 (0.56, 2.13) | 1.40 (0.72, 2.74) | 0.71 (0.59, 0.84) | |
| Pooled | 0.83 (0.54, 1.27) | 1.19 (0.79, 1.75) | 0.64 (0.52, 0.80) | |
| I2 | 11% | 0% | 82% | |
| MINI vs. SCID | PHQ-9 | 1.62 (1.05, 2.50) | 1.43 (0.91, 2.25) | 1.11 (1.00, 1.24) |
| EPDSc | 0.91 (0.43, 1.94) | 1.15 (0.52, 2.50) | 0.76 (0.62, 0.93) | |
| HADS-Dc | 1.52 (1.01, 2.30) | 1.57 (1.03, 2.40) | 0.96 (0.84, 1.09) | |
| Pooled | 1.46 (1.11, 1.92) | 1.45 (1.08, 1.93) | 0.95 (0.78, 1.15) | |
| I2 | 0% | 0% | 82% | |
| MINI vs. CIDI | PHQ-9d | 2.00 (1.13, 3.54) | 1.34 (0.75, 2.38) | 1.52 (1.37, 1.68) |
| EPDSc | 3.72 (1.21, 11.43) | 2.83 (0.85, 9.33) | 1.49 (1.18, 1.88) | |
| HADS-Dc | 1.70 (0.84, 3.43) | 1.40 (0.71, 2.76) | 1.34 (1.13, 1.58) | |
| Pooled | 2.05 (1.36, 2.10) | 1.49 (0.99, 2.25) | 1.48 (1.36, 1.60) | |
| I2 | 0% | 0% | 0% | |
No interaction; adjusted for depression symptom severity (standardized PHQ-9, EPDS, or HADS-D scores), age, and country human development index for all three IPDMAs, sex and patient care setting for the PHQ-9 and HADS-D IPDMAs, and pregnancy status (pregnant versus postpartum) for the EPDS.
Including an interaction between diagnostic interview and PHQ-9, EPDS, or HADS-D scores; adjusted for depression symptom severity (standardized PHQ-9, EPDS, or HADS-D scores), age, and country human development index for all three IPDMAs, sex and patient care setting for the PHQ-9 and HADS-D IPDMAs, and pregnancy status (pregnant versus postpartum) for the EPDS.
Results are slightly different from previously published results6,7 in terms of adjusted ORs for the interactions due to using standardized rather than raw scores in present analyses.
Only the two models of MINI vs. CIDI converged with the default optimizer in glmer, thus bobyqa was used instead for all other models.
Abbreviations: CIDI: Composite International Diagnostic Interview; EPDS: Edinburgh Postnatal Depression Scale; HADS-D: Depression subscale of Hospital Anxiety and Depression Scale; MINI: Mini International Neuropsychiatric Interview; PHQ-9: Patient Health Questionnaire-9; SCID: Structured Clinical Interview for DSM