Two Stage Design in a Community Survey

TLDR

The GHQ fails to fully differentiate between transient distress and true cases as defined by the PSE. In a two‑stage community survey, the GHQ‑60 performed well as a first‑stage screen, yet its specificity and positive predictive rates were low; increasing the cut‑off improved overall predictive values but sharply reduced sensitivity, and the revised scoring system did not outperform conventional scoring.

Abstract

By comparison with the PSE-ID system, we showed that the GHQ-60 could be used with good results as a screening instrument in the first of a two stage community survey. Unlike the sensitivity and negative predictive values, the specificity and positive predictive rates reach low figures in this study. The global efficiency and positive and negative predictive values of the test are significantly improved by raising its cut-off score, but at the expense of great reduction in sensitivity. The revised scoring system failed to provide better prediction of 'caseness' than conventional scoring. The GHQ does not distinguish fully between persons in the community afflicted by 'transient states of distress' and those whose symptoms would classify them as 'cases' by the PSE.