Prevalence of eating disorders among adolescent and young adult scholastic population in the region of Madrid (Spain)

Abstract

Objective

The prevalence of eating disorders (EDs) in a representative scholastic sample was evaluated, and the demographic factors associated with ED were assessed.

Method

The study used a two-stage epidemiological design. The study population was a randomly selected sample of 1545 students (12- to 21-year-old males and females) in the region of Madrid (Spain).

Results

ED prevalence was 3.43%. Prevalence estimations were as follows: 5.34% for females: 2.72% for eating disorders not otherwise specified (EDNOS), 2.29% for bulimia nervosa (BN), and 0.33% for anorexia nervosa (AN); and 0.64% for males: 0.48% for EDNOS, 0.16% for BN, and 0.00% for AN. Some demographic factors that were shown to be associated with ED included the following: sex, age, single child, single-parent families, and father or mother's death.

Discussion

Despite the methodological improvements over previous epidemiological studies developed in Spain, there were no significant differences in the prevalence estimations.

Introduction

In the last three decades, numerous epidemiological studies of the prevalence of eating disorders (EDs) have been developed in North America and Europe. Some of them have shown an incidence increase of two to five times since the 1960s to the 1970s [1], [2], [3], [4], [5], [6]. This increase is especially evident among groups that are at greater risk for EDs (i.e., adolescent and young women) [6], [7]. A very recent ED epidemiological study suggests, though, a decrease in bulimia nervosa (BN) prevalence in the last 20 years [8]. A recent review of the prevalence and incidence of ED [9] shows an average prevalence rate of 0.3% and 1% for anorexia nervosa (AN) and for BN, respectively, among young women. The estimated prevalence of BN for young men was 0.1%. The majority of the prevalence studies developed in other countries [9] have used a two-stage case-identification design that consists of introducing a screening questionnaire to identify the possible cases (i.e., those individuals who score above the cutoff point established in the questionnaire) and, afterward, interviewing those potential cases plus a random selection of controls using a clinical interview with diagnostic criteria for ED [e.g., from the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR) [10]].

The earliest epidemiological studies on ED in Spain were published in the first half of the 1990s. All of them have studied only scholastic populations. In spite of the advances in the research of these disorders, a detailed analysis of the ED prevalence studies published to date in Spain shows several methodological limitations.

For example, most of the previous studies are not proper ED prevalence studies. They do not actually include ED cases but, instead, identify risk behaviors or examine populations at risk [11], [12], [13], [14], [15], [16], [17], [18], [19], [20]. In addition, they do not use valid diagnostic instruments based upon diagnostic criteria for ED (like DSM-IV-TR [10]); instead, they rely upon screening questionnaires such as the Eating Attitudes Test-40 (EAT-40) [21], the Eating Disorders Inventory [22], or the General Health Questionnaire [23]. An isolated case [24] used a questionnaire with DSM-IV criteria of ED. These results thus cannot be considered to be prevalence data, given the limitations of the self-report questionnaires for purposes of case identification [25], [26], [27] (i.e., does not guarantee the correct interpretation of ED symptomatology, especially ambiguous concepts like “binges,” as a diagnostic interview does).

A second problem with previous Spanish studies, which has been emphasized in recent publications [28], [29], concerns their possible underestimation of the prevalence of EDs. Most of the double-stage prevalence studies [30], [31], [32], [33] interviewed only the individuals who scored equal to or above the cutoff point established for the screening instrument; therefore, it is not possible to determine whether false negatives were identified by the instruments administered. Only two studies [34], [35] interviewed some participants who scored lower than the cutoff point. The identification of false negatives is extremely important in the ED epidemiological studies, given that the affected subjects are predisposed to deny and hide their symptomatology.

A third problem concerns the use of the EAT-40, as a screening instrument, using a cutoff point of 30, as did the majority of the previous Spanish studies [11], [12], [15], [16], [19], [20], [32], [33], [34], [35], [36], [37], [38] following the original EAT-40 study [21], where a sensitivity of 100% and a specificity of 93% was found with this cutoff point. Nevertheless, further validation studies of the Spanish version of the EAT-40 [39], [40], using the score of 30 as the cutoff point, found a low sensitivity of 67.9% [39] and 75% [40] and a specificity of 85.9% [39] and 97.1% [40]. However, using the score of 20 as the cutoff point, the sensitivity was 91% and the specificity was 69.2% [39], and with a cutoff point of 25, the sensitivity was 87.5% and the specificity was 93.9% [40].

Considering that the majority of the Spanish two-stage epidemiological studies used the EAT-40 with a cutoff point of 30 and the fact that they did not interview participants whose scores were lower than 30, it is presumed that a percentage of ED cases was not detected in previous Spanish epidemiological studies.

A common characteristic of these studies, with some exceptions [41], is the absence of data about the association between ED cases and sociodemographic characteristics (e.g., single-parent families, sex, age, number of siblings, occupation of the parents, residence area, location within the region, and type of education center) or, in other words, the contributions or risks associated with these demographic variables in the development of ED.

These problems with the previous literature point toward a systematic bias (underestimating the prevalence of EDs) in the Spanish ED prevalence studies. Knowing the prevalence and risk of these disorders among scholastic general population is very important in planning and providing effectively the necessary mental health resources. Thus, trustworthy and precise ED epidemiological studies are needed. This kind of study must use a double-stage epidemiological design with a diagnostic interview in the second stage [25] to provide a more precise measure of ED pathology. Also, other methodological considerations such as the use (in the first phase) of a screening instrument with a low cutoff point (e.g., the Spanish version of the EAT-40 [39] with a cutoff point of 20) and interviewing several participants who score below the cutoff point of the screening instrument must be considered.