Steel et al ^{Reference Steel, Silove, Giao, Phan, Chey and Whelan1} should be commended for using an innovative design to show that the Composite International Diagnostic Interview (CIDI) 2.0 missed a large proportion of diagnoses that could instead be captured by an indigenously based Phan Vietnamese Psychiatric Scale (PVPS) among Vietnamese. Interpretations of the study should also consider the following.

1. (1) Comparison between the self-report PVPS and CIDI included two other methodological issues that have little to do with whether the PVPS was indigenously devised. First, face-to-face structured interviews have long been shown to bias against Asian populations in eliciting psychiatric symptoms. By contrast, Asian populations typically scored as high as Westerners on many self-report scales such as the General Health Questionnaire. ^{Reference Cheung2} Second, unlike the 53-item PVPS, the CIDI contains multiple skip-outs from further symptom questioning unless mandatory DSM–IV core symptoms are endorsed. This renders the hierarchically configured CIDI much more prone to false negatives. ^{Reference Lam, Pepper and Ryabchenko3}

2. (2) The majority of diagnoses captured by the PVPS (72%) were in the somatisation category, but somatoform disorders were not assessed in the CIDI (because of difficulty in operationalising the concept of ‘medically unexplained symptoms’). Recent versions of the CIDI (3.0 and 3.1) contain a section on chronic pains and other physical illnesses, which have been shown to be common and highly comorbid with mental disorders in both high-income and low- and middle-income countries. ^{Reference Scott, Von Korff, Alonso, Angermeyer, Bromet and Fayyard4}

3. (3) The CIDI surely requires improvement regarding downward bias in prevalence estimates in Asian countries. China has used several versions of it (1.0 to 3.1). By adhering strictly to linguistic accuracy, the earlier versions generated unbelievably low prevalence of depression. Prevalence estimates continue to rise with successive versions and the latest survey using CIDI–3.1, by taking careful account of contextual equivalence of stem questions, interviewer training and quality control in the field, has found a prevalence of depression little different from rates in many Western countries. The Chinese CIDI has also provided highly consistent epidemiological data regarding specific disorder distributions, lifetime rates, psychosocial associations, physical/mental comorbidity, treatment-seeking and the opportunity for large-sample cross-national analysis. ^{Reference Lee, Tsang, Von Korff, de Graaf, Benjet and Haro5} Enhancement of the CIDI may be both challenging and worth reconsidering in Vietnam.