In the present study, the validity of the Eating Disorder Inventory (€01) in a Dutch nonclinical female school population was established. The data were analyzed in two Ways. In the first analyses, the factorial integrity of the questionnaire, and the internal consistencies of the subscales were not found to be very satisfactory. The data showed an impressive lack of variability. It was suggested by the authors that in a nonclinical Population the variability ofthe ED1 data is being suppressed as a result of the fact that, In concordance with the manual, the original 6-point ED1 items were transformed into 4-Point items. In the second analyses, this item transformation was ejected. The psychometric properties of a// ED/ subscales in the present study improved. Implications for the use of the untransformed EDI as a screening instrument were discussed. 0 1994The Eating Disorder Inventory (EDI; Garner, Olmsted, & PoEvy, 1983) is a self-report questionnaire designed to measure psychological and behavioral traits commonly found u\ eating disorders. Although it was not designed for that purpose, it has been suggested as a screening instrument for eating disorders (Raciti & Norcross, 1987). There are two good reasons for this choice: Most of these psychological characteristics and behav-lorS measured by the ED1 are thought to proceed the onset of the illness, and secondly, the ED1 is the only instrument that provides assessment of a broad range of these variables. However, before a clinically derived measure can be used for early recognition of eating disorders in a nonclinical population, its validity in nonclinical populations must be well established (Hemchuk, Hutchinson, & Frank, 1990). For the EDI, especially CasWr Schoemaker is Research Assistant at the Department of Clinical Psychology at the University of Niimegen, where Tatjana van Strien, Ph.D., is staff member, and Cees van der Staak, Ph.D., is Professor. Correspondence should be addressed to the first author at Vakgroep Kfinische Psychologie en Persoonlijkheiddeer, Katholieke Universiteit Nijmegen, Postbus 9704, 6500 HE Nijmegen, The Netherlands.

The lack of communication and agreement by Dutch GPs and OPs in medical diagnosis and management of employees long-term sick listed due to mental health problems are indicators of sub-optimal medical treatment and return-to-work strategies.

Seasonal variations in the prevalence of mental disorders in general population surveys have rarely been noted. Using a representative sample of the Dutch population, the authors studied the seasonality of 1-month mental disorders and gender-by-season and age-by-season interactions. Data were obtained from the Netherlands Mental Health Survey and Incidence Study, a psychiatric epidemiologic study in which 7,076 adults aged 18-64 years were interviewed (1996) using the Composite International Diagnostic Interview. Seasonal statistical differences were not found for the category "any diagnosis", although prevalence was higher in winter (18.1%) and lower in summer (16.0%). Seasonal variations were also not found for the main categories "mood disorders", "anxiety disorders", "substance-use disorders", "eating disorders", and "schizophrenia". Some differences were found for individual disorders, mostly anxiety. Panic disorder was more frequently reported in winter than in summer, generalized anxiety disorder more frequently in winter than in other seasons, and obsessive-compulsive disorder more frequently in autumn than in summer. Only a few gender and age differences in seasonal variation were found. The authors conclude that there are only limited seasonal variations in mental disorders in general population studies, at least in countries with a mild maritime climate. For reliable estimation of the prevalence of some disorders, these findings have implications for the design of large-scale population studies.

Objective: Early intervention has been suggested to be an important positive prognostic factor in the treatment of anorexia nervosa (AN). The treatment-outcome studies conducted thus far were systematically reviewed in search of evidence to confirm this suggestion. Method: An extensive search was conducted for treatment outcome studies specifically addressing the prognostic value of the duration of the illness. Thirty-three follow-up studies of eating disorders and AN in particular were subjected to nine ''best-evidence'' criteria for methodological soundness. Six of these studies were found to meet all of the criteria and thus considered in greater detail. Results: In four of these studies, the duration of illness was not found to be related to the treatment outcome. This is presumably because of the low variability in duration of illness found in these studies. In the two remaining studies, a negative relation between duration of illness and the outcome of treatment was reported. As a result of methodological shortcomings and a lack of control for potentially confounding variables, however, the prognostic value of early intervention could still not be estimated on the basis of these studies. Discussion: The prognostic value of early intervention in AN cannot be estimated on the basis of the treatment-outcome literature. Unfortunately, the duration of the illness and not the occurrence of early intervention is considered in most of the outcome studies of AN. In addition, a number of methodological shortcomings make it impossible to evaluate the prognostic value of early intervention.

We found tentative support for the self-medication hypothesis.