[Interobserver variation of the Global Assessment Scale. Study of 59 patients at admission to a general adult psychiatric unit].

Unlabelled: In France, an analysis of the evaluation of psychiatric activity was done by a work group representing thirteen hospitals spread over the entire country. The group just recently published their final report in the form of an experimentation protocol for the description of psychiatric activity. The proposed methodology leads to some modifications of the existing data collection based on individual patient information sheets "fiche par patient", more particularly concerning new items such as the Global Assessment Scale (GAS or Axis V of DSM III-R): our first impression was that this new type of evaluation seemed to us somewhat subjective. We therefore decided to analyze the reliability of the score established by psychiatrists not familiar with its use. In the beginning the GAS score was evaluated in clinical cases (patients seen at their admission by two or three psychiatrists and scored afterwards independently), then in theoretical cases (34 cases types taken from the Health-Sickness Rating Scale of Luborsky) with the objective to improve the consensus between clinicians.

Results: The clinical case studies made in an adult psychiatry pavilion permitted evaluating 59 patients at admission, of which 25 by three judges and respectively 40, 28 and 41 by pairs. The global accuracy of the score obtained with the clinical cases seems satisfactory for a discipline such as psychiatry (Kappa of Cohen's coefficient = 0.51, p < 0.001 for 25 patients by three judges--Kappa = 0.45, p < 0.001 for the 109 pairs of evaluations). The more detailed study of the classes obtained a satisfactory consensus for the scores below 40 and above 60 (K = 0.46 to 0.59) but not entirely satisfactory for the scores in the mid range of the scale between 40 and 60 (K = 0.22 to 0.29) which represents 39% of the patients in the study; this is improved by grouping into larger classes this part of the scale and it is therefore advisable to use it that way. For the 34 theoretical cases taken from the Luborsky HSRS evaluated independently by three judges the results obtained are clearly not as good (global Kappa = 0.29, p < 0.01). That is perhaps due to the fact that these "case types" don't use the descriptive patient models generally used in France. Keeping in mind that the mid range of the scale should be interpreted prudently, the GAS scale can nevertheless be credited with a certain global objectivity in the case of neutral study. Would it be the same if the GAS study could influence the allocation of resources? Taking into account all the clinical practices in mental health, the question of how all the data of this type evaluation could be recorded and updated should also examined.