Nous avons réalisé une revue critique de la littérature afin de répondre à trois questions : (1) Faisons-nous face à une « épidémie » nationale et/ou internationale d’anorexie mentale (AM) comme le clament certains professionnels via les médias ? (2) L’AM n’est-elle qu’une banale crise d’adolescence ? (3) Est-ce une pathologie affectant exclusivement les catégories socioprofessionnelles élevées ?

Nous avons réalisé une recherche par Medline complétée par une recherche manuelle pour retrouver toutes les études ou revues de la littérature publiées sur la prévalence, le taux d’incidence, le devenir, la morbidité et la mortalité de l’AM durant la période 2000–2011.

Les différences de méthodologie des études menées sur ces thèmes ont des conséquences importantes sur l’interprétation des résultats. Pour chaque indicateur retenu, nous détaillons les résultats observés dans les différentes études et commentons à la lumière des différences méthodologiques.

Rien n’appuie l’idée d’une épidémie récente d’AM. L’AM n’est pas une simple crise d’adolescence compte tenu de la morbidité et de la mortalité qui sont considérables. La relation entre l’AM et la classe socio-économique n’est pas prouvée.

A critical review of the literature was conducted to provide answers to three questions: firstly, are we facing a national and/or international epidemic of anorexia nervosa (AN) as claimed in the media by certain professionals? Secondly, is AN simply an ordinary crisis of adolescence? Thirdly, is it a pathology that solely affects the higher socio-professional categories?

A Medline search was conducted and backed up by a manual search to find all the studies or literature reviews published on prevalence, incidence rates, outcomes, morbidity and mortality in AN and its links with social class, covering the period 2000–2011.

The differing methodologies of the studies on these themes have a considerable impact on the interpretation of results. For each indicator retained, the results observed in the different studies are detailed and commented in the light of these differences. The prevalence of AN in women aged 11 to 65 in non-clinical population ranges from 0 to 2.2 %. It varies in particular with the age of the studied subjects, the measurements used, and the definition criteria for AN. Among men, the lifetime prevalence is around 0.3%. The marked disparities in incidence rates observed are related to the nature of the samples considered: subjects consulting in hospital, whether in a specialised department or in any department, those consulting general practitioners (GPs), or subjects from general population samples recruited in different surveys (girls in schools for example). The incidence of female cases is low in general medicine or specialised consultation in town (whatever the speciality): from 4.2 and 8.3/100,000 individuals per year. It is much higher in the general population, ranging from 109 to 270/100,000 individuals per year. In fact, the studies reporting variations in the incidence of AN were conducted on samples from clinical populations in certain countries (United States and United Kingdom). They are probably more a reflection of variations in detection rates and use of healthcare, than of variations in the incidence in the general population. The mean duration of AN appears shorter in the general population than in clinical populations. On average, 47% of the individuals treated for AN recovered, 34% improved, 21% had a chronic eating disorder, and 5% died. The outcome is better for subjects treated during adolescence. Mortality is frequently expressed in crude mortality rate (CMR), which is not very informative on account of the heterogeneous natures of the cohorts followed; only the studies reporting standardised mortality rate (SMR) are informative. AN appears as having one of the highest mortality rates among psychiatric pathologies. Mortality varies according to the population considered. Rates observed are 6.2 to 10.6 times greater than that observed in the general population for a follow-up duration ranging respectively from 13 to 10years. It is lower for longer follow-up periods, only 3.7 times more frequent than in the general population for follow-up periods of 20 to 40years. It appears lower for subjects treated before the age of 20. The main causes of death are eating disorder complications, suicide and cancer. One review of the literature concluded in the absence of any significant link between this pathology and social class.

There is nothing in the incidence and prevalence data to back up the notion of a recent “epidemic” of AN. AN is not simply a crisis of adolescence: morbidity and mortality are considerable in this pathology. The relationship between AN and social class is not established.