Using quantitative measures Turner et al (Reference Turner, Bowie and Dunn2003) found that about half of a sample of 842 Kosovan refugees in the UK had post-traumatic stress disorder, with substantial comorbid depressive disorder and anxiety disorder. But there is more to be reported. I was involved in having a few open-ended questions tacked on to the study, tapping subjects’ own views of their health/mental health and what they saw as their most urgent priorities for recovery. Only a tiny number saw themselves as having a mental health problem of any kind, bearing out observations by refugee workers in the reception centres housing them that there was no interest in counselling. Almost everyone nominated work, schooling and family reunion as their major concerns. This chimes with what I and others have found in clinical settings with refugees over many years. Significant psychopathology is uncommon (Reference SummerfieldSummerfield, 2002).

The responses to the open-ended questions paint a picture that is a world away from that reported by Turner and colleagues; how is this contradiction to be explained? First, the question of validity. Translation/back-translation of psychiatric inventories originating in the USA and Western Europe does not by itself overcome the category fallacy to which Kleinman (Reference Kleinman1987) pointed: particular phenomena may be identified in different settings but it does not follow that they mean the same thing in each setting. Moreover, refugees in distressed and insecure circumstances may be particularly susceptible to the demand characteristics of questionnaires. Second, and fundamentally, how human beings experience an adverse event, and what they say and do about it, is primarily a function of the social meanings and understandings attached to it. No psychiatric category captures this active appraisal and meaning-making.

Quantitative methodologies serving psychiatric categorisations risk a distorting pathologisation of refugee distress, with what is social and collective being reassigned as individual and biological (Reference SummerfieldSummerfield, 1999). Turner et al caution against ‘the tendency of some to reject the diagnostic paradigm in refugee populations’, but they do not make a persuasive case here that they know better than the Kosovan refugees themselves, and that many of the refugees really do need psychiatric treatment. There is simply no good evidence to back their conclusion that refugee populations anywhere are carrying a major burden of clinically significant mental ill health. As the answers to my questions demonstrated, refugees see recovery as primarily something that must happen in their social worlds, not in the space between their ears.