Dückers et al ^{Reference Dückers, Alisic and Brewin1} analyse the relationship between prevalence estimates of trauma exposure and post-traumatic stress disorder (PTSD) in published data-sets from 24 countries, and between PTSD and vulnerability (based on a country vulnerability index developed in the 2013 World Risk report). The findings are substantially counterintuitive; countries with low vulnerability have higher life-time rates of PTSD, meaning that countries with low vulnerability and high trauma exposure have the highest rates of lifetime PTSD. The authors do emphasise a number of limitations of their work, and yet they conclude that a ‘vulnerability paradox’ exists for both PTSD and depression, with rates higher in countries with more resources and better healthcare systems.

This conclusion would seem inconsistent with a great deal of work in global mental health, which emphasises the considerable treatment gap in mental health services, with under-diagnosis and under-treatment particularly high in low- and middle-income countries. ^{Reference Susser and Patel2} It raises the question of what precisely is being measured by epidemiological studies of common mental disorders in general, and by studies of trauma exposure and PTSD in particular. ^{Reference Patel, Maj, Flisher, De Silva, Koschorke and Prince3} There has been no shortage of critics of psychiatric nosology, including the construct of PTSD: ^{Reference Stein, Seedat, Iversen and Wessely4} are counterintuitive findings such as those of Dückers et al valid in some way, or do they underscore the limitations of our current classification systems, and the epidemiological surveys which employ related measures?

Consider, for example, the findings cited by Dückers et al that in South Africa and Lebanon, 73.8% and 68.85% of the population reported exposure to trauma, lower rates than in The Netherlands or Canada. In our view, given the multiple influences that determine self-reported rates of trauma exposure (including those noted by Dückers et al), comparing such rates across surveys is a matter of ‘comparing oranges and apples’. Other data from other sources may legitimately allow comparison of prevalence estimates: for example, the death rate from motor vehicle accidents in South Africa is 25.1 per 100 000 compared with 3.4 in The Netherlands, and there were 35.7 v. 8.9 murders per 100 000 in South Africa v. The Netherlands. ⁵ Furthermore, rigorous examination of raw data across surveys (which Dückers et al note that they did not undertake) allows valid conclusions about trauma exposure: for example, that a small number of traumatic events account for a larger proportion of all traumatic event exposure across the world. ⁵

When it comes to PTSD, Dückers et al note a prevalence of PTSD of 0.0% in Nigeria, 3.4% in Lebanon, and 9.2% in Canada; they emphasise a range of methodological issues that may have contributed to such findings, but nevertheless proceed to their analysis. In our view, the 0.0% prevalence estimate of PTSD in Nigeria should be considered as a single sampling, prone to any number of measurement errors. ^{Reference Benjet, Bromet, Karam, Kessler, McLaughlin and Ruscio6} While many sociocultural factors may affect the prevalence of PTSD, given the many universal findings about the phenomenology and psychobiology of PTSD, ^{Reference Yehuda, Hoge, McFarlane, Vermetten, Lanius and Nievergelt7} drawing strong inferences from this single data-point is not a scientifically sound approach. Again, however, other data and other analytic approaches do allow rigorous conclusions regarding the cross-national epidemiology of PTSD. For example, rigorous analysis of raw data from the World Mental Health Surveys has indicated that dissociative symptoms indicate a particularly severe and impairing subtype of PTSD. ^{Reference Stein, Koenen, Friedman, Hill, McLaughlin and Petukhova8}

In view of these considerations, we wish to express our scepticism about the construct of a ‘vulnerability paradox’. This is not to criticise all of the fascinating literature on health paradoxes; it may well be the case that well and well-informed populations complain more about health problems than ill but ill-informed populations. ^{Reference Barsky9} And it is not to ignore the considerable methodological issues facing psychiatric classification and epidemiology in general, as well as particular issues relevant to trauma such as the causal relationship between trauma exposure and a range of disorders other than PTSD. ^{Reference Bryant, O'Donnell, Creamer, McFarlane, Clark and Silove10} Instead, our argument is that given these issues, certain kinds of analyses (such as those undertaken by Dückers et al) are fundamentally flawed, and the field should instead focus on those analyses which allow rigorous conclusions about trauma exposure and PSTD.