In summary, our report identified lower diagnostic concordance between the CIDI–2.0 and the indigenously derived PVPS among Vietnamese in the Mekong Delta region compared with Vietnamese in Australia. Whereas rates of mental disorder identified by the PVPS were stable across countries, the CIDI-identified mental disorder was three times lower in the Mekong Delta. Of particular importance was that the CIDI failed to detect 75% of Vietnamese with similar levels of disability identified by the PVPS.

Lee et al raise important questions that need to be resolved in order to make sense of the findings of international psychiatric epidemiology. We address some of their concerns in relation to our method. Although technically the PVPS is a questionnaire, it was administered in interview format as is common in the transcultural setting. Moreover, there is some evidence that among Vietnamese, there is a tendency to use a restricted range in reporting symptom severity on questionnaires, ^{Reference Beiser and Fleming1} a factor that would yield conservative rates. Lee et al suggest that the skip rules of the CIDI may lower prevalence rates. We concur that the pre-eminence given to psychological rather than somatic stem symptoms in the hierarchical structure of the CIDI ^{Reference Shen, Zhang, Huang, He, Liu and Cheng2} might limit positive endorsements in non-Western countries. However, if this effect was present it differentially had an impact on the Mekong Delta sample, underscoring the importance of culture and ‘Westernisation’ as an influence on psychiatric assessment. We look forward to the publication of the results from the Chinese trials of the CIDI–3.1, which have reformulated the stem questions to be more compatible with somatic idioms of distress.

We do note, however, that removing PVPS cases that only reached threshold on the somatisation scale would have reduced our prevalence rates by 2.8% in Vietnam and 3.0% in Australia. Hence, the PVPS would still have identified a substantial number of cases not yielded by the CIDI. We note too that the Western-derived measure of neurasthenia recorded low rates in all samples, suggesting that somatic measures need to be culture specific.

In summary, there does not seem to be any major disagreement here. Whether we produce indigenous measures ab initio, as we have done, or modify existing measures as undertaken by Lee et al with the CIDI–3.1, the inference we draw remains the same: in order to detect the full range of disabling mental disorders across cultures, we need to have culturally appropriate measurements. We cannot simply apply the same measure with the same wording of items in the same format to all cultures and expect that we can compare the results. The cost of applying either adapted or culturally developed measures, however, is that it confounds the process of making direct international comparisons of prevalence rates and mental health need. Hence, the real challenge facing world psychiatry is how to combine the strengths of psychiatric epidemiology ^{Reference Kessler, Angermeyer, Anthony, de Graaf, Demyttenaere and Gasquet3} with improvements in culturally valid assessment. ^{Reference Silove, Bateman, Brooks, Zulmira, Steel and Rodger4,Reference Patel, Simbine, Soares, Weiss, Wheeler and Patel5} Showing consistent patterns of comorbidity and risk-factor profiles across countries can only partially address this issue.