Prevalence of eating disorders in middle-aged women


  • Supported by the Public Health Services of the City of Innsbruck (Austria).


Objective: Little is known about the prevalence and correlates of eating disorders (ED) in middle-aged women. Method: We mailed anonymous questionnaires to 1,500 Austrian women aged 40–60 years, assessing ED (defined by DSM-IV), subthreshold ED, body image, and quality of life. We broadly defined “subthreshold ED” by the presence of either (1) binge eating with loss of control or (2) purging behavior, without requiring any of the other usual DSM-IV criteria for frequency or severity of these symptoms. Results: Of the 715 (48%) responders, 33 [4.6%; 95% confidence interval (CI): 3.3–6.4%] reported symptoms meeting full DSM-IV criteria for an ED [bulimia nervosa = 10; binge eating disorder = 11; eating disorder not otherwise specified (EDNOS) = 12]. None displayed anorexia nervosa. Another 34 women (4.8%; CI: 3.4–6.6%) displayed subthreshold ED. These women showed levels of associated psychopathology virtually equal to the women with full-syndrome diagnoses. Discussion: ED appear common in middle-aged women, with a preponderance of binge eating disorder and EDNOS diagnoses as compared to the “classical” diagnoses of anorexia and bulimia nervosa. Interestingly, middle-aged women with even very broadly defined subthreshold ED showed distress and impairment comparable to women with full-scale ED. © 2013 Wiley Periodicals, Inc. (Int J Eat Disord 2014; 47:320–324)


Most research on eating disorders (ED) and body-image concerns has focused on girls and young women, but recent surveys have begun to suggest that ED are not uncommon among individuals at midlife or beyond.[1-3] In accord with this finding, recent general-population surveys have found that the mean age of individuals reporting eating-disorder behaviors is relatively high.[4, 5] Data regarding body image in aging women remain sparse, but generally suggest that body dissatisfaction and drive for thinness do not diminish with age.[6, 7] Our clinical experience with older women has echoed these findings and has further suggested that even women with seemingly mild or infrequent eating-disorder behaviors may nevertheless report substantial associated psychopathology. To test whether these clinical impressions would hold in a non-clinical sample, we assessed eating behavior and body image in 715 community women aged 40–60 in Innsbruck, Austria.



We have previously detailed our survey design in a preliminary report[8] examining menopausal status and ED in a subgroup of the study respondents. Here, we briefly reiterate the design, followed by an analysis of the full group of study respondents.

Using the mail, we invited 1,500 randomly selected women in Innsbruck, Austria, aged 40–60 years, to complete an anonymous questionnaire about ED and related symptoms. Of these 715 (48%) returned evaluable questionnaires. The questionnaire included demographic items, three subscales of the Eating Disorder Inventory (EDI),[9] questions from the German version of the Structured Clinical Interview for DSM-IV (SKID),[10] the Body Shape Questionnaire (BSQ),[11] the Diagnostic Survey for Eating Disorders (DSED),[12] the Center for Epidemiologic Studies Depression Scale (CES-D),[13] and the World Health Organization Quality of Life-BREF (WHOQOL-BREF).[14] Details of these instruments, including references to articles describing their psychometric properties, are provided in our previous article.[8]

We diagnosed current ED by DSM-IV criteria, based on answers to the SKID questions, as detailed previously.[8] In addition, as mentioned above, we hypothesized that even women with limited eating-disorder symptoms might nevertheless exhibit significant pathology. Accordingly, we diagnosed SED if respondents reported simply (1) binge eating accompanied by loss of control or (2) purging behaviors including use of laxatives, diuretics, appetite suppressants, and/or vomiting, but not including excessive exercise or extreme dietary restriction—regardless of the frequency or severity of these symptoms (note technically that misuse of appetite suppressants is not listed as an example of purging in DSM-IV, but that misuse of “other medications” is now included as an example of purging in DSM-5). Thus, our SED category was much broader than traditional criteria for “subthreshold” ED, which typically fall only slightly below full DSM-IV eating-disorder criteria.[15] Women meeting neither ED nor SED criteria were defined as “normal eaters” (NE).

Statistical Analyses

We compared the ED, SED, and NE groups using the Kruskal-Wallis test for continuous variables and chi-square for nominal variables. Post-hoc pair-wise group comparisons were performed using the Mann-Whitney U test and Fisher's exact test, two-tailed. We adjusted for body-mass index (BMI) in comparisons involving body-image measures. We performed post-hoc comparisons only if the overall three-group comparison was significant (p < .05)—a procedure permitting the alpha-level of 0.05 to be retained without correction for multiple testing. We have detailed these procedures previously.[8]


Among the 715 respondents, 90.1% listed Austria as their birth country; 75.2% were married or living with a partner, and 76.5% had children—figures closely comparable to those for similarly aged women in the Austrian general population.[16] However, respondents were better educated than Austrian women overall (41.6% reporting ≥ 12 years of education vs. 18–26% of general-population women).[16]

Thirty-three respondents [4.6%; 95% confidence interval (CI): 3.3–6.4%] reported symptoms meeting DSM-IV criteria for a current ED (anorexia nervosa = 0, bulimia nervosa = 10, and EDNOS = 23). Of the EDNOS cases, 11 displayed binge eating disorder (now an official diagnosis in DSM-5, and hence no longer subsumed under EDNOS); six met full DSM-IV criteria for anorexia nervosa, but lacked amenorrhea (N = 3) or reported weight in the normal range despite significant weight loss (N = 3); and six reported regular inappropriate purging behavior, despite normal body weight, after eating small amounts of food (laxative abuse = 2, diuretic abuse = 3, and appetite suppressants = 1).

Thirty-four women (4.8%; CI: 3.4–6.6%) met our criteria for SED. Of these, 19 reported eating binges with loss of control, but failed to meet the DSM-IV frequency or severity criteria for full-scale binge eating disorder. The remaining 15 women reported purging after eating small amounts of food (self-induced vomiting = 1, laxative abuse = 7, diuretic abuse = 3, appetite suppressants = 4), but were overweight, thus not qualifying for EDNOS, which is diagnosed only for purging in the presence of normal weight.

The ED, SED, and NE groups showed no significant difference in age distribution, but the ED group displayed significantly lower educational attainment and more frequent non-European origin compared to the ED group (Table 1).

Table 1. Demographic and clinical features of women with eating disorders, subthreshold eating disorders, and normal eating
 GroupSignificance of Differencesa
N = 33 (5%)N = 34 (5%)N = 648 (90%)OverallED vs. SEDED vs. NESED vs. NE
  1. a

    See text for detailed statistical methods.

  2. b

    ED—eating disorders (DSM-IV).

  3. c

    SED—subthreshold eating disorders (see text for definition).

  4. d

    NE—normal eating.

  5. e

    EDI—Eating Disorder Inventory.

  6. f

    BMI—Body Mass Index (based on self-reported height and weight).

  7. g

    CES-D—Community Epidemiologic Scale for Depression.

  8. h

    WHOQOL-BREF—WHO quality of life.

Age: exactly below 40–49 years, N (%)18 (55)23 (68)368 (57)ns   
50–60 years, N (%)15 (46)11 (32)275 (43)    
Married or in partnership, N (%)20 (63)27 (82)486 (76)ns   
Children (> 1), N (%)26 (79)26 (77)493 (77)ns   
Education > 12 years, N (%)6 (19)9 (27)272 (42)0.007ns0.009ns
Non-European origin, N (%)4 (12)1 (3)12 (2)0.001ns0.006ns
EDIe—drive for thinness, M (SD)9.5 (5.4)7.6 (5.9)2.4 (3.6)0.000ns0.0000.000
EDI—bulimia3.5 (3.6)1.7 (2.6)0.2 (0.8)0.0000.0460.0000.000
EDI—body dissatisfaction18.7 (5.4)18.9 (6.2)11.2 (5.5)0.000ns0.0000.000
EDI—total 3 subscales31.1 (11.5)27.6 (10.7)13.7 (8.4)0.000ns0.0000.000
BMIf current, M (SD)26.7 (5.3)27.8 (6.4)23.7 (4.3)0.000ns0.0010.000
BMI desired22.9 (2.6)23.8 (3.0)22.0 (2.5)0.000ns0.0390.000
Lifetime restrictive dieting, N (%)   0.000ns0.0000.002
Often-Very Often (> 20 lifetime diets)8 (26)5 (17)18 (3)    
Never-Sometimes (0–20 lifetime diets)23 (74)24 (83)600 (97)    
CES-Dg Total score, M (SD)22.4 (10.9)18.8 (11.3)10.8 (8.8)0.000ns0.0000.000
Clinical depression (cutoff > 16), N (%)23 (74)19 (58)139 (22)0.000ns0.0000.000
Psychological domain55.6 (18.2)60.3 (19.9)74.3 (15.1)0.000ns0.0000.000
Physical domain67.1 (20.2)70.0 (18.9)82.6 (14.2)0.000ns0.0000.000
Social domain62.4 (20.2)68.4 (21.0)72.9 (19.0)0.009ns0.005ns
Environment domain69.1 (15.6)74.8 (13.3)80.1 (13.2)0.000ns0.0000.017
Global domain57.0 (22.2)64.4 (23.0)76.6 (18.4)0.000ns0.0000.001

Associated Features

On a wide variety of measures, the women with ED reported substantial associated pathology (Table 1). Compared to the women with normal eating, both the ED and the SED groups demonstrated significantly higher scores on the EDI subscales and total score, higher current and desired BMI, more frequent self-reported lifetime dieting to lose weight, higher CES-D depression scores, and higher scores on every domain of the WHOQOL-BREF. On the BSQ and various subjective questions regarding body image, both eating-disorder groups again differed significantly from the NE group across the board (Table 2). Interestingly, across all measures, the ED and SED groups showed virtually no significant differences from one another, despite the much broader criteria for the latter group.

Table 2. Body image measures in women with eating disorders, subthreshold eating disorders and normal eating
 GroupSignificance of Differencesa
N = 33 (5%)N = 34 (5%)N = 648 (90%)OverallED vs. SEDED vs. NESED vs. NE
  1. a

    See text for detailed statistical methods.

  2. b

    ED—eating disorders (DSM-IV).

  3. c

    SED—subthreshold eating disorders.

  4. d

    NE—normal eating.

  5. e

    BSQ—Body Shape Questionnaire.

How important is your appearance? N (%)   ns   
Very important24 (73)21 (62)423 (66)    
Moderately important8 (24)13 (38)14 (33)    
Not important1 (3)09 (2)    
How satisfied are you with your weight? N (%)   0.000ns0.0000.000
Very important3 (9)2 (6)303 (48)    
Moderately important12 (36)10 (29)201 (31)    
Not important18 (55)22 (65)136 (21)    
How satisfied are you with your shape? N (%)   0.000ns0.0000.000
Very important3 (9)1 (3)292 (45)    
Moderately important14 (42)11 (33)242 (38)    
Not important16 (49)21 (64)111 (17)    
How fat do you feel? N (%)   0.000ns0.0000.000
Very fat9 (28)8 (24)50 (8)    
Moderately fat18 (56)24 (71)222 (34)    
Not at all fat5 (16)2 (6)374 (58)    
Self evaluation is influenced by weight and shape, N (%)28 (88)22 (67)280 (44)0.0000.0760.0000.012
BSQ scoree, M (SD)119 (33)104 (34)68 (28)0.000ns0.0000.000
“I really like my body!” N (%)   0.000ns0.0000.000
Agree6 (19)10 (29)441 (69)    
Unsure12 (38)13 (38)154 (24)    
Disagree14 (44)11 (32)47 (7)    


Using mailed questionnaires, we assessed the prevalence of ED and associated symptoms in a community sample of 715 women age 40–60 in Innsbruck, Austria. The study yielded three principal findings. First, our results augment the growing evidence[1-5] that ED are common in middle-aged women. Notably, the eating-disordered group showed lower educational attainment and more frequent non-European origin than NE, consistent with prior reports on the “democratization” of ED.[17] Second, middle-aged women showed a preponderance of binge eating disorder and eating disorder not otherwise specified (EDNOS) as compared to the “classical” diagnoses of anorexia and bulimia nervosa. Some clinicians and investigators may be less familiar with EDNOS presentations of ED and hence less likely to detect them.

Third, we found that women with even very broadly defined SED showed virtually the same levels of associated pathology as women with full-scale DSM-IV ED. This finding appears consistent with recent observations in younger women.[18, 19] Normally, such subthreshold cases would receive no DSM-IV diagnosis at all and might be missed in clinical or research settings.

Several limitations of the study should be recognized. First, only 48% of the initial sample of 1,500 women provided evaluable responses, and we lacked information on non-respondents. Although respondents resembled the Austrian general population on several demographic indices, they were better educated than Austrians overall. Other types of selection bias may also have occurred. For example, since ED are often secret, participants with ED may have been less likely to respond than those without, causing us to underestimate the true rates. Conversely, we might have overestimated the true prevalence of behaviors such as binge eating, if individuals with this behavior were more likely to respond or if they overstated the severity of binge-eating behavior on our self-report instrument. Second, we diagnosed ED by DSM-IV criteria using self-report questions derived from the SKID, an interview normally administered verbally. Although this approach has previously been used in another questionnaire survey,[3] it has not been formally tested for reliability and validity. Also, since our SKID-derived questions were specifically keyed to DSM-IV, we were unable to reliably re-diagnose respondents using DSM-5 criteria. Third, the questionnaire did not allow us to determine the age of onset, and hence the degree of chronicity, of the reported current ED. One prior Australian survey,[20] covering both sexes and a wider age range (age 15–95; mean 46.9 years), reported a mean (SD) duration of 6.6 (9.2) years for eating-disorder behaviors, suggesting a wide variation in chronicity.

Our findings have several implications for clinicians and researchers. First, ED appear common in middle-aged women, and many such cases might be missed if clinicians are not expecting them and inquiring about them. Second, women with even a single eating-disorder symptom, who would normally fall well below the threshold for a formal diagnosis, may nevertheless suffer substantial distress. If this finding were supported by subsequent studies, it would argue for expanded diagnostic criteria to recognize such cases in some manner lest they be ignored. It would also follow that clinicians, when learning that a patient exhibits uncontrolled binge eating or purging, even in the absence of other diagnostic features of an ED, should inquire carefully about associated pathology, such as depression and body-image concerns. Further research will be invaluable to expand our limited understanding of these issues.