Nous avons réalisé une revue de la littérature critique évaluant la prévalence des troubles de l’humeur chez les sujets présentant un trouble du comportement alimentaire (anorexie mentale = AN ou boulimie = BN). Dans la première partie, nous avons discuté les problèmes méthodologiques posés par ces études. Dans la deuxième partie, nous avons pris en compte ces problèmes méthodologiques et avons résumé les résultats. Nous avons réalisé une recherche manuelle et une recherche par Medline pour retrouver toutes les études publiées sur la comorbidité entre troubles du comportement alimentaire et troubles de l’humeur. Nous avons limité notre recherche à la période 1985-2002, afin de retenir les études ayant utilisé les critères diagnostiques stricts pour les deux types de troubles du comportement alimentaire. Beaucoup d’études n’ont pas inclus de groupe témoin et très peu ont comparé la prévalence des troubles de l’humeur dans les différents sous-groupes de troubles du comportement alimentaire. Les résultats sont, dans les diverses études, peu fiables ou contradictoires. Nous avons discuté les résultats décrits et les implications de cette revue de la littérature pour la réalisation d’études futures.

Objective – Comorbidity between eating disorders (ED) and mood disorders is a major issue when eva­luating and treating patients with anorexia nervosa (AN) or bulimia nervosa (BN). In the literature, estimated comorbidity rates of mood disorders in subjects with ED differ widely across studies. Obviously, it is difficult to compare results from various sources because of differences in methods of assessment of depressive symptoms and in diagnostic criteria for both ED and mood disorders. Furthermore, few studies have included control groups, and, since mood disorders are among the most frequent psychiatric disorders in women – with an average estimated lifetime prevalence of 23.9 % (Kessler et al., 1994) –, it is not clear, yet, whether mood disorders are more common among women with an ED (AN or BN) than among women from the community. The only review articles we found on the relationships between ED and mood disorders survey different types of arguments in favour of a link between both categories of disorders, including symptoms, personal and family comorbidity, overlap in biological findings, and treatment results, but do not review in detail available comorbidity data. The aim of this paper is to conduct a critical literature review on studies assessing the prevalence of mood disorders in subjects with an ED (AN or BN). In the first part, we will discuss methodological issues relevant to comorbidity studies between ED and mood disorders, and select the most reliable studies. In the second part, taking into account these methodological considerations raised, we summarize the findings of these studies. Method – We performed a manual and computerized search (Medline) for all published studies on comorbidity between ED and AD, limi­ting our search to the 1985-2002 period, in order to get sufficiently homogeneous diagnostic criteria for both categories of disorders. Results – Too few studies include control groups and few studies have compared diagnostic subgroups of ED subjects, with scarce or conflicting results. Discussion – We reviewed numerous studies here and conclude simply that there are many arguments in favor of elevated rates of MD in ED subjects, but there is no convincing evidence yet. Many questions are left unanswered or have conflicting responses. Our review highlights the need for further studies, which should address several requisites : comorbidity studies should be designed with this as a specific goal, rather than as a secondary aim within other types of studies (such as treatment studies, follow-up studies, etc.). Kendler et al. (1991) state that individuals with two disorders are more likely to present for treatment than individuals with one, therefore, comorbidity rates (which are not in agreement with a special etiologic relationship between BN and depression) may be exaggerated in clinical population results. New studies should include control subjects, matched (at least) for sex and age with ED subjects. Studies should evaluate prevalence of all types of MD in order to yield comparable estimates of MD in general. Comorbidity studies should be conducted on both current and recovered patients, compared to subjects from the community. It is still necessary to demonstrate specificity of findings, i.e. that early onset MD are of specific etiological importance to ED and do not simply increase the risk of later psychopathology in general. Studies should be conducted on larger samples, and all diagnostic subgroups should be considered (restrictive and bulimic anorexics, bulimics with and without history of AN, with or without purging). Multivariate comparisons should be performed, taking into account subject age, sex (if men are included), in- and outpatient status, course of illness, and other possibly relevant variables. Thus, more reliable estimates of the frequency of MD in subjects with ED could provide us with valuable etiologic, therapeutic and prognostic information.