N J Shelton1, K G Herrick. 1. Department of Epidemiology and Public Health, University College London (UCL), 1-19 Torrington Place, London WC1E 6BT, UK. n.shelton@ucl.ac.uk
Abstract
OBJECTIVES: To compare the scoring methods and thresholds of the 12-item General Health Questionnaire (GHQ-12) and the Edinburgh Postnatal Depression Scale (EPDS) in English women, and to determine which threshold and scoring method provides the closest correlation of caseness of postnatal depression in a nationally representative sample of English women. STUDY DESIGN: Health Survey for England 2002 health examination survey. METHODS: Self-completion booklet containing the EPDS and the GHQ-12. Participants were mothers with at least one child under 1 year of age at the time of interview. RESULTS: Both the scoring method and cut-off affected the estimates of prevalence of postnatal depression in English women. The best threshold and scoring method for the GHQ-12 using sensitivity/specificity analysis against the EPDS was a standard scale with a cut-off of 3+. This matched the cut-off using the GHQ-12 mean scores. The cut-off using comparative prevalence of the GHQ-12 with the EPDS was higher at 4+. There was a significantly lower estimate of prevalence of postnatal depression at 4 months using the GHQ-12. CONCLUSIONS: Care needs to be taken measuring postnatal depression. The GHQ-12 mean score cut-off matched the cut-off using sensitivity and specificity; this supports using the GHQ-12 mean scores as cut-offs. The standard scale was most closely correlated with the EPDS. Although there was strong correlation between the GHQ-12 and the EPDS, a significantly lower proportion of women were measured as having possible postnatal depression at 4 months using the GHQ-12. This may be due to the lack of a question on blame in the GHQ-12. Four months coincided with the duration of maternity leave entitlement and recommended age for weaning in 2002. These events may be particularly stressful for mothers, and practitioners need to be mindful of similar milestones for diagnosis if using the GHQ-12.
OBJECTIVES: To compare the scoring methods and thresholds of the 12-item General Health Questionnaire (GHQ-12) and the Edinburgh Postnatal Depression Scale (EPDS) in English women, and to determine which threshold and scoring method provides the closest correlation of caseness of postnatal depression in a nationally representative sample of English women. STUDY DESIGN: Health Survey for England 2002 health examination survey. METHODS: Self-completion booklet containing the EPDS and the GHQ-12. Participants were mothers with at least one child under 1 year of age at the time of interview. RESULTS: Both the scoring method and cut-off affected the estimates of prevalence of postnatal depression in English women. The best threshold and scoring method for the GHQ-12 using sensitivity/specificity analysis against the EPDS was a standard scale with a cut-off of 3+. This matched the cut-off using the GHQ-12 mean scores. The cut-off using comparative prevalence of the GHQ-12 with the EPDS was higher at 4+. There was a significantly lower estimate of prevalence of postnatal depression at 4 months using the GHQ-12. CONCLUSIONS: Care needs to be taken measuring postnatal depression. The GHQ-12 mean score cut-off matched the cut-off using sensitivity and specificity; this supports using the GHQ-12 mean scores as cut-offs. The standard scale was most closely correlated with the EPDS. Although there was strong correlation between the GHQ-12 and the EPDS, a significantly lower proportion of women were measured as having possible postnatal depression at 4 months using the GHQ-12. This may be due to the lack of a question on blame in the GHQ-12. Four months coincided with the duration of maternity leave entitlement and recommended age for weaning in 2002. These events may be particularly stressful for mothers, and practitioners need to be mindful of similar milestones for diagnosis if using the GHQ-12.