The concept of prevalence applied to psychiatric disorders and symptoms.

Statements about prevalence, of the type 'x% of the general population are mentally ill', imply a concept of prevalence that is difficult to define operationally. It requires the specification of a cut-off point, a level of severity from which the presence of an illness can be inferred, and this is largely arbitrary. A more generally useful type of statement compares the frequency distribution of severity for declared cases with that for the rest of the population. This avoids the need for a defined cut-off point. We are still left with the problem of defining a dimension of severity and a declared case, but these should follow easily from a clear statement of the purpose of any specific inquiry. The argument is illustrated from the results of a study of self-referrals to general practitioners.