| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Neurology and Psychiatry at the University of Padova, Padova, Italy.
Address reprint requests to: Dott.ssa Silvia Ferrara, Clinica Psichiatrica, Dip. Scienze Neurologiche e Psichiatriche, Via Giustiniani 3, 35128 Padova, Italy.
Received for publication August 2, 2002; revision received October 7, 2002.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
METHOD: All female subjects aged 18 to 25 who resided in two areas (urban and suburban) of a large city were involved in the study. All women (N = 934) underwent a clinical interview which included the structured clinical interview for DSM-IV
RESULTS: Lifetime anorexia nervosa (AN) and bulimia nervosa (BN) were diagnosed respectively in 2.0% and 4.6% of the subjects. The prevalence of lifetime atypical ED was 4.7% and that of binge eating disorder (BED) was 0.6%. The degree of urbanization has a significant impact on the prevalence of AN, BN, and BED. Social class, professional status, and education were not associated with an increased risk of reporting an ED, whereas the number of hypocaloric diets, having been a victim of childhood abuse, and, in BN, ever being overweight are significantly associated with ED.
CONCLUSIONS: Our findings have confirmed the importance of community studies to improve our knowledge about factors that have some influence on pathogenesis, treatment referral, and outcome.
Key Words: anorexia nervosa, • bulimia nervosa, • prevalence, • childhood abuse, • binge eating disorder, • eating disorders.
Abbreviations: AN = anorexia nervosa;; BED = binge eating disorder;; BMI = body mass index;; BN = bulimia nervosa;; DSM-IV = Diagnostic and Statistical Manual of Mental Disorders, fourth edition;; ED = eating disorders;; PED = purging eating disorder;; SCID = structured clinical interview for DSM-IV.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Community studies have substantially confirmed the DSM-IV classification of full-syndrome ED (11, 12). However, it is still a controversial issue on how to consider atypical ED, ie, those patients who do not fulfill all the criteria specified by the DSM-IV for AN or BN (13). Generally, those subjects who display all the symptoms of a full syndrome, but do not reach the AN weight threshold or frequency of binge eating episodes in BN, are defined as subthreshold ED. On the contrary, subjects completely lacking one of the diagnostic criteria are considered partial ED (for example, AN subjects without amenorrhea or BN subjects without objective binges or compensatory behavior). The general population study by Garfinkel and colleagues (9) has suggested that partial AN (ie, AN in subjects who do not fulfill the criterion of amenorrhea) does not differ from full AN as regards weight loss, history of sexual abuse, comorbidity rates, and family psychiatric history. Similarly, they have demonstrated that subjects with BN do not differ from subjects who fulfill all the criteria of BN except for a binge eating frequency of once a week (7). This issue was also analyzed from a genetic-epidemiological point of view by Sullivan and colleagues (14) using a large sample of twins. They found that the risk ratio to the co-twin of the same definition of binging reaches a peak at four objective binges per month and not at the DSM-IV threshold of twice per week.
Studies performed on general population samples using cluster analysis (11) or latent class analysis (12, 14) reported controversial results as regards the classification of atypical eating disorders. Sullivan et al. (14) have confirmed the existence of binge eating disorder, whereas others suggested caution in the formulation of this "new" diagnosis (11).
These studies are not of particular help in describing other atypical ED, because they tend to group subjects who share a symptomatic dimension, though with different degrees of severity (12). The findings of these studies tend to support the enlargement of diagnostic criteria but do not provide data about the epidemiology and clinical importance of atypical ED.
Few studies to date have examined the prevalence of atypical ED. Little is known about the natural history of these disorders and the rate at which they shift toward a full syndrome ED (15, 16) . The aim of the present study is to evaluate the current and lifetime prevalence of the whole spectrum of eating disorders in a female general population sample using the DSM-IV criteria.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
All women aged 18 to 25 years who were listed on these registers were sent a letter containing the aims and methods of the research study and an invitation to participate. In the space of a few weeks, every subject was then contacted by phone to arrange for their clinical interview. All subjects were interviewed in a face-to-face fashion by the second author (S.F.), who had received specific training for conducting interviews in the field of eating disorders. At least three phone contacts were programmed for all subjects; those not traceable after these attempts were visited at home. All participants gave written informed consent for the use of data in an anonymous form. Clinical interviews lasted 30 to 60 minutes and were carried out within a period of 12 months.
There were 1306 female subjects aged 18 to 25 who resided in the two areas. Of these, 108 had temporarily or permanently changed their address and were excluded (total sample = 1198). We were unable to trace 34 (2.8%) of the remaining subjects and another 230 (19.2%) refused to be interviewed. A final response rate of 78% was achieved, resulting in 934 interviewed subjects. The two areas did not differ as regards response rate (76.5% vs. 79.4%). Information about refusers was obtained by consulting their general practitioner and the archives of the eating disorders unit of Padova University (the only public ED unit in the area).
Measures and Definitions
Clinical interviews were performed using the ED Section of the structured clinical interview for DSM-IV (SCID) (17). Subjects answered all the key questions in the interview as well as the screening questions. Some other semi-structured questions were administered concerning age, educational degree, social class, marital and professional status, age at menarche, the highest and lowest weight after menarche, the number of hypocaloric diets, and history of sexual and/or physical abuse. Childhood abuse was defined as the history of either sexual or physical abuse occurring before the age of 18. Social class was determined using an Italian adaptation of Havighurst’s formula (18). This formula calculates social class using paternal and maternal professional status and degree of education.
Anorexia nervosa, bulimia nervosa, and binge eating disorder were diagnosed using the DSM-IV criteria (19). Partial syndrome ED are defined as follows: Atypical AN is defined as a) partial AN when all of the criteria for AN are met except that individuals never develop amenorrhea or b) subthreshold AN when all of the criteria for AN are met except that, despite significant weight loss, the individual’s weight remains in the normal range. Atypical BN is defined as a) partial BN when all of the criteria for BN are met except that individuals reported recurrent episodes of subjective (as opposed to objective) binge eating and regular use of inappropriate compensatory behavior of the nonpurging type; b) subthreshold BN when all of the criteria for BN are met except that binge eating and inappropriate compensatory behavior occur at least twice a month but less than twice a week; c) purging eating disorder (PED) when all of the criteria for BN purging type are met except that objective binge eating is not present; or d) binge eating disorder (BED).
All diagnoses, except BED (which requires a period of 6 months), must be present for a period of at least 3 months. It is difficult to assess how many subjects with lifetime full ED also satisfied all the criteria for atypical ED before the development (or in a phase of remission) of full ED. For this reason, in the calculation of the lifetime prevalence of atypical AN, we excluded subjects with lifetime AN and, in that of atypical BN, we excluded subjects with lifetime BN. On the contrary, the point prevalence of atypical ED may include subjects who have had full ED.
Objective binge eating was defined as "eating an amount of food that is definitely larger than most individuals would eat under similar circumstances" (19). When the amount of food was not definitely large or the interviewer was uncertain, the binge eating was defined as subjective. Subjects were considered "behaviorally remitted" when they did not display any of the following symptoms in the 3 months before the interview: low body weight (BMI < 18.5), amenorrhea, objective binge eating, self-induced vomiting, abuse of laxatives and/or diuretics, fasting, severe restraint, or food avoidance.
Statistics
SPSS software was used. The t test, one-way analysis of variance, and 
2 (with Yates’ continuity correction for 2 x 2 tables) were performed. Confidence intervals of prevalence were calculated with the following formula: p ± 1.96 (p x q/N)1/2; where p is the prevalence (n/N), q is (1-p), n is the number of affected subjects, and N is the total number of subjects. The statistical significance of the differences between prevalences was calculated using the formula proposed by Robins et al. (20): t =(p1 - p2)/[(p1 x q1/N1) + (p2 x q2/N2)]1/2. The relationships between social and clinical factors and case status were assessed by the 2 likelihood ratio statistic. In addition, odds ratios with their 95% confidence intervals were reported. The significance of odds ratios was assessed by the Mantel-Haenszel test and the Breslow-Day test was used to test the null hypothesis of homogeneity of the odds ratio in two or more groups.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
Among subjects with lifetime AN, 10 (53%) were in complete remission at the time of the interview. Four subjects with lifetime AN restricting type (29%) and two with AN binge eating/purging type (40%) shifted to a diagnosis of bulimia nervosa (five BN-P and one BN-NP). One subject with lifetime restricting AN shifted to a subthreshold BN that lasted about 6 months before a full remission; another shifted to partial BN (still present) and two others to partial AN, the diagnosis at the time of the interview.
The point prevalence of bulimia nervosa (BN) was 1.8% (17 cases) whereas the lifetime prevalence was 4.6% (43 cases). Twenty-nine percent of the currently affected subjects (N = 5) and 56% of the subjects with a lifetime BN (N = 24) met the criteria for purging type BN (the lifetime ratio between the two subtypes is about 1:1). Thirteen subjects (30%) had received or were still receiving some form of treatment for ED, but only six (14%) had been referred for psychological or psychiatric treatment.
Five purging (21%) and one nonpurging bulimics (5%) had had a previous episode of AN. Five purging (21%) and three nonpurging bulimic subjects (16%) reported a previous episode of subthreshold AN (amenorrhea after an important loss of weight) before the onset of BN, and two subjects with nonpurging BN (10%) reported a previous episode of partial AN (BMI lower than threshold without amenorrhea). Apart from the lifetime lowest BMI, which was significantly lower among bulimics with previous AN (15.8 ± 1.3 vs. 19.5 ± 1.8, t = 4.74, p < .001), no differences emerged in the age of onset (19.0 ± 3.5 vs. 16.8 ± 3.0, t = 1.61, NS) and duration of binge eating (38.0 ± 22.0 vs. 39.9 ± 37.5 months, t = 0.12, NS) between BN subjects with previous AN and those without. Seven purging (29%) and five nonpurging BN subjects (26%) reported the occurrence of a previous overweight condition (BMI > 25). Three subjects with previous BN and three subjects with current BN were overweight at the time of the interview. Subjects with a lifetime overweight condition did not differ from those without as regards age of onset and duration of binge eating. No statistically significant difference emerged between purging and nonpurging BN as regards age, age of onset, duration of illness, or lifetime lowest and highest BMI.
Among subjects with lifetime BN, 19 (44%) did not meet the criteria for any form of ED in the 3 months before the interview. Seven cases of purging BN (29%) had shifted to an atypical form of ED (four subthreshold BN and three purging ED).
From the consulting general practitioners and the archives of the Padova ED unit, we found two cases of AN (one case with restricting AN and one case with binge eating/purging AN) and a case of purging BN among the subjects who refused the interview.
Atypical ED
In the definition of atypical AN, we included subjects with partial AN (N = 15) and those with subthreshold AN (N = 12). Table 3 reports the main characteristics of the two groups compared with the subjects with full AN. Three cases of partial AN were not included in Table 1, nor in the lifetime rate of atypical AN, because they have lifetime AN. The point and lifetime prevalence of atypical AN was 0.7% (7 cases) and 2.6% (24 cases), respectively. Five cases of lifetime subthreshold AN shifted to a diagnosis of purging BN, three others to a diagnosis of nonpurging BN, and two cases of lifetime partial AN shifted to nonpurging BN. One subject with nonpurging BN changed to partial AN (BMI = 15.1 without binge eating and amenorrhea) but later returned to a diagnosis of nonpurging BN. None of the other subjects with atypical AN reported having any other form of ED before its onset.
|
), because they reported a diagnosis of lifetime full BN. The point and lifetime prevalence of binge eating disorder is 0.1 (1 case) and 0.6 (6 cases), respectively, and the point and lifetime prevalence of the other types of atypical bulimia is 2.4% (22 cases) and 3.1% (29 cases), respectively.
|
Referring to the archives of the Padova ED unit, we found a new case of BED among the subjects who refused the interview.
Differences Between the Urban and the Suburban Area
As expected, the subjects of the urban area are significantly different than those of the suburban area as regards many of the sociodemographic variables (Table 1). In the urban area, fewer women aged 18 to 25 are married but they have a higher educational level. In addition, 53.7% of women in the urban area and 74.7% of those in the suburban area belong to the medium-low or low social class. The two samples differ also as regards the distribution of age: in the urban area women aged 22 to 25 are more prevalent than women aged 18 to 21. No difference between the two samples emerged as regards BMI (20.7 ± 2.6 vs. 21.1 ± 3.2).
There was a higher lifetime prevalence of ED in the urban area compared with the suburban one (12.8% vs. 9.1%, t = 1.74, p < .05). Table 5 shows the lifetime prevalences of the full and partial ED in the two areas. The lifetime prevalences of AN, BN, and BED are significantly higher in the urban sample. If we consider the diagnostic subgroups, the difference between the two areas is greater for the nonpurging BN subgroup (3.1% vs. 1.0%, t = 2.19, p < .02) than for the purging BN subjects (3.1% vs. 2.1%, t = 0.97, NS). Both restricting AN (2.0% vs. 1.0%, t = 1.18, NS) and binge eating/purging type AN (0.9% vs. 0.2%, t = 1.39, NS) tend to be more prevalent in the urban area.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Although some authors have described an overrepresentation of ED cases among subjects who choose not to participate in surveys on ED (27), our findings, obtained by referring to general practitioners and to the archives of our ED unit, did not seem to confirm this observation. However, because treatment referral is not high among ED subjects, some other cases were probably present and it is possible that our prevalence figures may underestimate the actual prevalence of ED in the community.
Atypical ED
Our study reveals that about 11% of the women in our sample suffer or have suffered from an ED. Atypical ED do not seem to be more frequent than full ED. This is in contrast with most of the literature (28), but the reason may lie in the methodology of our study. First of all, we used strict diagnostic criteria for the definition of atypical ED which included a minimum duration of the symptoms of at least 3 months. Second, we excluded from the diagnosis of lifetime atypical ED all subjects who later developed full ED and those who displayed atypical ED during the remission of full ED. Finally, though the period of time studied is relatively short, it is possible that the reliability of reporting past atypical ED is lower than that of reporting full ED (29). If we consider only those studies which employed strict and precise diagnostic criteria, the prevalences of atypical ED vary from study to study. For atypical AN, the reported point prevalence ranges from 0.5% (24) to 1.0% (9) and even to about 3% (30, 31). For atypical BN, the prevalence ranges from 1.2% (7, 30) to about 3% (24, 25).
All the different forms of atypical ED we considered are reported with a lifetime prevalence ranging from 0.6 to 1.3%. Garfinkel et al. found that 1% of their female general population sample suffer from partial AN (9) and 0.5% from subthreshold BN (7). These figures are very similar to ours, 1.3% and 0.9%, respectively. Few prevalence studies to date have evaluated the prevalence of BED. In the study performed in an urban area of Southern Italy (22), the point prevalence of BED was 0.2%. The literature suggests that, unlike other ED, BED is not typical of adolescence. The mean age of onset among BED subjects recruited in a clinical setting is usually significantly higher than in other ED (32). This is probably why we found a low prevalence of BED in our sample. Epidemiological studies with a broader age range are needed to obtain reliable prevalence figures for this diagnosis.
No available study to date has evaluated the frequency of subthreshold AN, partial BN, and purging ED. Although the current BMI of subthreshold AN was in the normal range, they lost a great deal of weight (Table 3) and are very similar to AN subjects as regards body attitudes and fear of gaining weight. Partial BN, on the contrary, is perhaps the most questionable subgroup of atypical ED because of the absence of both binging and purging behavior. Like PED, it is described in DSM-IV as the fourth example of ED not otherwise specified. We have maintained a distinction between partial BN and PED because, in our clinical experience, PED subjects tend to become chronic, are less motivated to treatment, and have a high frequency of medical complications (33). This clinical impression was not confirmed in the present study (except for two cases where PED occurred after a diagnosis of BN: in both the purging behavior was still present and lasted about 9 years). It would seem that, in the general female population, this subtype of ED varies greatly as regards the degree of severity but, probably, only the most severe cases are referred for treatment.
Characteristics of Full ED
Our study gives a cross-sectional view of the whole spectrum of ED as they are present in the young adult female general population. This viewpoint is important because it is well known that only a minority of subjects with ED seek treatment. A deeper knowledge of the characteristics of ED in the general population leads us to make some hypotheses about the factors that can influence treatment referral. In our study, we observed that AN with binge eating/purging features represents a minority of all subjects with lifetime AN (26%). Despite its low prevalence, this subgroup probably represents the ED group with highest costs for the public health system due to the greater likelihood of medical and psychiatric complications, poorer outcomes, and higher risk of mortality (34, 35). These factors probably explain why, in clinical samples of young women with ED, this group often consists of more than 20% of the subjects. In BN, we found that the purging and nonpurging subgroups have a very similar prevalence, whereas in clinical samples purging BN usually represents the great majority of BN subjects (6). This means that nonpurging BN subjects ask for treatment less often than purging BN. Indeed, none of the subjects with nonpurging BN in our sample reported having asked for psychiatric or psychological treatment whereas 18% of purging BN did so. If we consider that the reporting of nonpurging BN is probably less reliable than that of purging BN (29), we may conclude that the prevalence of nonpurging BN is equal to or even higher than that of purging BN.
It is noteworthy that the present study found a high frequency of subjects with lifetime full AN who show a behavioral remission, in some cases spontaneous, of their symptoms (Table 3). This result is in contrast with the accepted view that AN tends to display a chronic course. Our study is limited by the fact that, being cross-sectional, it is not known how many subjects will have a relapse. At the time of the interview, remitted subjects were normal weight (mean BMI = 19.5 ± 1.1) and had regular menses, although in some cases they maintained a restrictive attitude toward food (40%) and fear of gaining weight (30%). The duration of the illness among remitted AN was 14.4 months on average (range = 6–24 months), with 70% of cases having sought some type of treatment. These findings might lead us to hypothesize that there are subjects in the general population who develop a more benign form of AN without a chronic course which, sometimes, remits without requiring specific treatment. Although fewer bulimics showed complete remission (Table 4), this might also apply to BN. Remitted subjects who have had BN were completely abstinent from binge eating after a duration of illness of about 3 years (range = 3–120 months). Only 37% of them asked for some type of treatment and in about half of the cases they still reported restrictive attitudes toward food (47%) and an excessive influence of body weight and shape on self-esteem (47%). These findings are interesting but have to be considered with caution because we need more information about these cases and their general health. Greater knowledge about behaviorally remitted cases in the general population would allow us to better understand the mechanisms of chronicity and to evaluate the effectiveness of specific treatments.
Other findings in this study would seem to be of clinical interest. We found that 21% of purging BN and 5% of nonpurging BN had had a previous episode of AN (14% of all subjects with lifetime BN). These rates are slightly higher than those reported by Kendler et al.(25), who found 10% of previous AN, but consistently lower than those reported for clinical samples (25–37%) (14, 36). However, if we consider also the episodes of atypical AN, the rates of BN subjects with previous full or atypical AN is 42% for the purging and 32% for the nonpurging subgroup.
Degree of Urbanization
The generalizability of our prevalence rates is limited by the fact that we did not assess a sample recruited in a rural area. Because we found a significant difference in the prevalence of AN and BN between urban and suburban areas, it might well be that the prevalence of these disorders in rural areas is lower than that found in the present study.
The finding of a higher prevalence of disorders in areas with a greater degree of urbanization is not new in psychiatric literature. However, this is the first prevalence study to examine the impact of this factor on the frequency of ED. Hoek and colleagues (23) found a significant difference in the incidence rates of BN, but not AN, among areas with a different degree of urbanization. They hypothesized that the role of social pressure to be slender (which is greater in cities) could play a more important role in the pathogenesis of BN than in that of AN, whose origin might be more closely linked to biological and genetic factors. Moreover, other factors, such as the different age distribution, the fact that people are more likely to live alone in large cities, and the greater availability of food, may be influential (23).
In our sample, unlike in Hoek’s study, the relationship between epidemiological rates and degree of urbanization seems to regard all types of ED, with the exception of atypical diagnoses. However, in comparing our findings with those of Hoek’s study, we should bear in mind that prevalence rates are influenced not only by factors important in the pathogenesis but also by those that affect the maintenance of a disorder. Different prevalences of psychiatric disorders in areas with different degrees of urbanization have been observed in disorders for which the role of genetic factors is unquestionable, such as schizophrenia (37) or major depression (38). Indeed, the presence of genetic vulnerability (39) does not exclude the influence of social factors on the prevalence of a psychiatric disorder.
The contrast between the urban and suburban areas might be linked to the social differences we observed in the two areas (Table 1). This hypothesis has not been confirmed by the statistical analyses reported in Table 6. Social class, education, age, and professional and marital status did not show any significant association with ED diagnoses. Ever being overweight seems to increase the risk of BN but it seems to be slightly more frequent in the suburban area (18% vs. 14%). Finally, the association of ED with the number of hypocaloric diets and with physical and childhood abuse was the same in the two areas, showing no effect of urbanization on these factors.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. Ahren-Moonga, R. Silverwood, B. a. Klinteberg, and I. Koupil Association of Higher Parental and Grandparental Education and Higher School Grades With Risk of Hospitalization for Eating Disorders in Females: The Uppsala Birth Cohort Multigenerational Study Am. J. Epidemiol., September 1, 2009; 170(5): 566 - 575. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. T. EDDY Dr. Eddy Replies Am J Psychiatry, June 1, 2008; 165(6): 773 - 773. [Full Text] [PDF]
|
|
|
|
|
|
A. FAVARO and P. SANTONASTASO The Value of Anorexia Nervosa Subtypes Am J Psychiatry, June 1, 2008; 165(6): 772 - 773. [Full Text] [PDF]
|
|
|
|
 |
|
 A. Favaro, E. Tenconi, and P. Santonastaso The Relationship Between Obstetric Complications and Temperament in Eating Disorders: A Mediation Hypothesis Psychosom Med, April 1, 2008; 70(3): 372 - 377. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. D. Wade, M. Tiggemann, C. M. Bulik, C. G. Fairburn, N. R. Wray, and N. G. Martin Shared Temperament Risk Factors for Anorexia Nervosa: A Twin Study Psychosom Med, February 1, 2008; 70(2): 239 - 244. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Favaro, E. Tenconi, and P. Santonastaso Perinatal Factors and the Risk of Developing Anorexia Nervosa and Bulimia Nervosa Arch Gen Psychiatry, January 1, 2006; 63(1): 82 - 88. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
