How can we be sure that the diagnosis of schizophrenia actually relates to a valid entity? In Chapter 18, van Os and Verdoux challenge the existing beliefs about schizophrenia as a discrete diagnostic category and consider the evidence that psychosis exists as a continuous and dimensional phenotype. There is accumulating evidence that psychosis exists in nature as a continuous distribution of symptoms rather than a discrete, dichotomous disorder. Large-scale national surveys have found that about one-quarter of individuals surveyed admit to experiencing psychotic symptoms and most had never sought any help for these. In addition, there is evidence that psychosis consists of many overlapping symptom dimensions that may each be the result of a range of underlying risk factors. Van Os and Verdoux argue that a combination of categorical, continuous and multidimensional representations of psychosis will offer important advantages both for clinical practice and research.

The final two chapters in the book demonstrate how epidemiology can be used to improve the treatment of schizophrenia and, perhaps, ultimately to prevent the disorder. In Chapter 19, Thornicroft and Tansella select epidemiological findings that have been discussed in earlier chapters and, using a ‘matrix’ model, they show how these data can be exploited to plan better services for people who suffer from schizophrenia.