A two-phase psychiatric survey of a random sample of 18-64 aged respondents from Nijmegen Health Area is described. In phase I 3232 persons answered the Dutch version of the GHQ. 485 of them, some at random chosen, some selected because of a GHQ score of ten or higher, were then interviewed with the PSE. This resulted in 101 cases, defined as having an Index of Definition of five or higher.
In the article we discuss GHQ and PSE separately, their use for studying a general population and a logistic model representing the relationship of PSEcaseness to GHQ scores.
We found an amazing similarity with the logit formule, calculated by Henderson e.a. (Australia) indicating a general applicability of the model, independent of age and sex, for computing prevalences, given only the frequence distribution of GHQ scores for a population. For the Nijmegen Health Area population the prevalence of PSE-cases was 7,3%.
The consequences of the model for calculating sensitivity and specificity of the GHQ as a screeningtest are discussed.