Screening for postnatal depression in C hinese‐speaking women using the H ong K ong translated version of the E dinburgh P ostnatal D epression S cale

The E dinburgh P ostnatal D epression S cale ( EPDS ) may not be a uniformly valid postnatal depression ( PND ) screen across populations. We evaluated the performance of a C hinese translation of 10‐item ( HK‐EPDS ) and six‐item ( HK‐EPDS ‐6) versions in post‐partum women in S ingapore. Methods C hinese‐speaking post‐partum obstetric clinic patients were recruited for this study. They completed the HK‐EPDS , from which we derived the six‐item HK‐EPDS ‐6. All women were clinically assessed for PND based on D iagnostic and S tatistical M anual, F ourth E dition– T ext R evision criteria. Receiver–operator curve ( ROC ) analyses and likelihood ratio computations informed scale cutoff choices. Clinical fitness was judged by thresholds for internal consistency [α ≥ 0.70] and for diagnostic performance by true‐positive rate (>85%), false‐positive rate (≤10%), positive likelihood ratio (>1), negative likelihood ratio (<0.2), area under the ROC curve ( AUC , ≥90%) and effect size (≥0.80). Results Based on clinical interview, prevalence of PND was 6.2% in 487 post‐partum women. HK‐EPDS internal consistency was 0.84. At 13 or more cutoff, the true‐positive rate was 86.7%, false‐positive rate 3.3%, positive likelihood ratio 26.4, negative likelihood ratio 0.14, AUC 94.4% and effect size 0.81. For the HK‐EPDS‐6, internal consistency was 0.76. At 8 or more cutoff, we found a true‐positive rate of 86.7%, false‐positive rate 6.6%, positive likelihood ratio 13.2, negative likelihood ration 0.14, AUC 92.9% and effect size 0.98. Discussion The HK‐EPDS (cutoff ≥13) and HK‐EPDS6 (cutoff ≥8) are fit for PND screening for general population post‐partum women. The brief six‐item version appears to be clinically suitable for quick screening in C hinese speaking women.