In the European Prediction of Psychosis Study (EPOS)N = 37 participants converted to psychosis, n = 146 completed the whole 18-month follow-up period without conversion. Assessed by the Global Assessment of Functioning Scale, modified version (GAF-M), the following functional states were considered: Considered GAF-M: ¡Ü30 % /¡Ü10 % reduction of baseline scores related to highest scores in the previous year; scores ¡Ü70/¡Ü60.
The GAF reduction criteria led to a very unfavorable loss of sensitivity, even, if only 10 % were demanded. This was accompanied by correspondingly unfavorable accuracy measures. Introducing functional impairment criteria defined by the current state reported to be predictive for psychiatric caseness (score ¡Ü 70) or to define serious impairment (score ¡Ü 60) (Kessler et al., 2002, 2003) kept sensitivity at a perfectly high level, yet did not produce any gain of specificity.
These results were certainly be caused by the fact that the whole group showed already low GAF-M scores in the previous year. Thus, a functional impairment criterion proved not to be useful to improve prediction. However, a combination of APS or BLIPS with a ¡®clinical status¡¯ criterion of GAF-M ¡Ü 70 may be considerable to demonstrate a strong need for intervention.