Published on in Vol 14, No 5 (2012): Sep-Oct

Disordered Eating in a Digital Age: Eating Behaviors, Health, and Quality of Life in Users of Websites With Pro-Eating Disorder Content

Disordered Eating in a Digital Age: Eating Behaviors, Health, and Quality of Life in Users of Websites With Pro-Eating Disorder Content

Disordered Eating in a Digital Age: Eating Behaviors, Health, and Quality of Life in Users of Websites With Pro-Eating Disorder Content

Original Paper

1The Craig Dalsimer Division of Adolescent Medicine, Department of Pediatrics, The Children's Hospital of Philadelphia, Perelman School of Medicine at The University of Pennsylvania, Philadelphia, PA, United States

2Division of Pediatric Neurology, Department of Pediatrics, The Children's Hospital of Philadelphia, Perelman School of Medicine at The University of Pennsylvania, Philadelphia, PA, United States

3Department of Pediatrics, Stanford University, Stanford, CA, United States

4Department of Neurology, Children's National Medical Center, Washington, DC, United States

5Division of Child and Adolescent Psychiatry, Department of Psychiatry and Behavioral Sciences, Stanford University School of Medicine, Palo Alto, CA, United States

6Rice University, Houston, TX, United States

7Department of Health, Behavior and Society, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, United States

Corresponding Author:

Rebecka Peebles, MD

The Craig Dalsimer Division of Adolescent Medicine

Department of Pediatrics

The Children's Hospital of Philadelphia, Perelman School of Medicine at The University of Pennsylvania

3401 Civic Center Blvd

11NW Room 19

Philadelphia, PA, 19104

United States

Phone: 1 215 590 6864

Fax:1 215 590 4708


Background: Much concern has been raised over pro-eating disorder (pro-ED) website communities, but little quantitative research has been conducted on these websites and their users.

Objective: To examine associations between levels of pro-ED website usage, disordered eating behaviors, and quality of life.

Methods: We conducted a cross-sectional, Internet-based survey of adult pro-ED website users. Main outcomes were Eating Disorder Examination Questionnaire (EDE-Q) and Eating Disorder Quality of Life (EDQOL) scores.

Results: We included responses from 1291 participants; 1254 (97.13%) participants were female. Participants had an average age of 22.0 years and a mean body mass index of 22.1 kg/m2; 24.83% (296/1192) were underweight; 20.89% (249/1192) were overweight or obese. Over 70% of participants had purged, binged, or used laxatives to control their weight; only 12.91% (163/1263) were in treatment. Mean EDE-Q scores were above the 90th percentile and mean EDQOL scores were in the severely impaired range. When compared with moderate and light usage, heavy pro-ED website usage was associated with higher EDE-Q global (4.89 vs 4.56 for medium and 4.0 for light usage, P < .001) and EDQOL total scores (1.64 vs 1.45 for medium and 1.25 for light usage, P < .001), and more extreme weight loss behaviors and harmful post-website usage activities. In a multivariate model, the level of pro-ED website usage remained a significant predictor of EDE-Q scores.

Conclusions: Pro-ED website visitors reported many disordered eating behaviors, although few had been treated. Heavy users reported poorer quality of life and more disordered eating behaviors.

J Med Internet Res 2012;14(5):e148



The Internet offers numerous websites that can affect the health of vulnerable users. Of particular concern are pro-eating disorder (pro-ED) website communities (also called pro-anorexia or pro-ana, and pro-bulimia or pro-mia), where individuals may learn about, discuss, and reinforce disordered eating behaviors [1]. Pro-recovery websites promote discussion more related to fighting an eating disorder, although online content can overlap between pro-recovery and pro-ED communities [2].

Content on pro-ED websites includes “thinspiration” (images or text for the purpose of inspiring weight loss), techniques to assist in weight loss, and interactive forums [2-6]. Some sites promote eating disorders as a lifestyle choice, offering encouragement for extreme dieting and exercise behaviors and assistance in avoiding detection by family and medical providers. Other websites aim to support visitors at various stages of illness, recognizing the dangers of disordered eating and offering content dedicated to treatment and recovery [4]. See Figure 1 for an example of a mock website.

These websites can have deleterious effects on the user [7-10]. Participants exposed to a pro-ED site for 25 minutes were more likely to show negative affect, perceive themselves as heavier, and exercise or think about weight [11]. Pro-ED visitors have also displayed higher body dissatisfaction, restriction, and bulimic activity than controls [12]. A pilot study of 76 adolescents who had been in treatment for an eating disorder found that over a third had visited pro-ED sites. Practically all (96%) of these pro-ED website users reported learning new weight loss or purging techniques from the sites [1].

Great concern has been raised over pro-ED websites; however, relatively little is known about their users. No study has examined a large group of pro-ED website users from a clinical perspective, nor have associations between disordered eating patterns and escalating levels of site visitation been described. The purpose of this study was to examine the demographics, media use patterns, and eating behaviors of pro-ED website visitors, and the degree to which website usage correlates with disease severity and quality of life.

Figure 1. Mock pro-eating disorder website with typical content.
View this figure

We recruited adult users of websites with pro-ED content via a dedicated link established on identified websites. All protocols were approved by the Stanford University Panel on Medical Research in Human Subjects, Stanford, CA, USA, and were compliant with the Health Insurance Portability and Accountability Act of 1996. Informed consent was obtained from all participants.

Website Search and Inclusion

We developed a comprehensive new methodology to access users from a broad spectrum of websites with pro-ED content. We searched Google and Yahoo!, entering the keywords “pro-ana” and “pro-mia”. All 700 URL addresses on the first 35 pages of search results for either engine were examined. Websites were included if they displayed any of the following: (1) a declaration that the website was pro-ED, (2) a disclaimer or warning to stay away from the site if the visitor was in recovery or did not have an eating disorder, (3) the term thinspiration, or (4) information on disordered eating behaviors in a framework intended to inform the disorder (tips and techniques). We further searched these websites for links to similar sites. These second-generation websites were also examined and included if they met the above criteria. Additionally, we searched the three most popular open social networking sites as determined by comscore Media Metrix [13] for the keywords pro-ana, ana, pro-mia, and mia. Webrings (collections of related websites) or interest groups found in these searches were included if they met the inclusion criteria and had at least 500 members.

Our methods involved no deception. Only English-language sites with active participation or updates within the last 12 months were included. Websites or webrings were excluded if their maintainers indicated that they were less than 18 years of age or did not wish to be involved in research. Figure 2 details the search results.

Figure 2. Website search results. a First 35 pages of results included. b Groups with 500 or more members included. c Included in this category are pro-ED (eating disorder) sites under construction with little ED content, pro-ED sites no longer containing content because maintainer went into recovery, sites with primarily non-ED-related content, and sites containing only links to other sites.
View this figure


We contacted the maintainers of 296 websites, describing the study and its purpose, and asking maintainers to post a link to our study website, which described both the study and the online survey. Participation time was described as 30–40 minutes. On entering the study site, potential participants were asked to confirm that they were over 18 years of age and to give informed consent before being directed to the survey. No incentives were used to encourage participation. No identifiers were collected and no cookies were used. IP addresses were not tracked or limited in their responses. Participants were able to contact the study researchers via email. Survey responses were entered online and automatically captured into a data file.


Our online survey contained 193 items, covering demographics, self-reported heights and weights, disordered eating, quality of life, treatment history, overall health, self-harm, and website usage of both pro-ED and pro-recovery websites (see Multimedia Appendix 1). Participants were asked to self-designate their race or ethnicity according to categories defined by the US National Institutes of Health. The Eating Disorder Examination Questionnaire (EDE-Q) assessed the degree of disordered eating and thoughts [14], the core set of the US Centers for Disease Control and Prevention’s (CDC) Health-Related Quality of Life Healthy Days Measure [15] was used as a generic quality-of-life measure, and the Eating Disorder Quality of Life (EDQOL) measure [16] assessed eating disorder-specific health-related quality of life. Finally, we adapted a survey that our team had previously used to evaluate pro-ED and pro-recovery website usage [1]. Participants were also asked if they lied on the survey, and if so, these responses were excluded from analyses. The survey was piloted within all members of the research team to be certain it was functional online prior to fielding the questionnaire.

Outcome Measures

Main outcome measures were scores on the EDE-Q and EDQOL. Secondary outcomes were the number of unhealthy days on the Healthy Days Measure and common indicators of health in patients with an eating disorder, such as history of admission to hospitalization, osteopenia, or missed menses. The predictor variable for all primary analyses was the level of pro-ED website usage.

Weight Calculations

We calculated body mass index (BMI) using participants’ self-reported heights and weights, using the equation BMI = weight in kilograms/(height in meters)2. Reported BMI values were divided by a median BMI of 21.7 for women and 23.0 for men, based on growth charts developed by the CDC [17], to obtain percentages of median body weight for each participant. Participants were asked for their highest and lowest weights and ages at those times, and definitions of overweight and obesity at those times followed CDC guidelines for children and adults.


We used standard descriptive and frequency testing to describe the data. Associations were assessed using chi-square testing, Student t test, Pearson correlations, and analysis of variance, followed by Tukey test for post hoc comparisons. Multivariate regression analyses were used to stratify factors predictive of disordered eating behaviors and impaired quality of life. Responses were not weighted in any manner. Significance level was set at .05. Statistical analysis was conducted using SPSS 19.0 (IBM Corporation, Somers, NY, USA).

Response Rate

While nonresponse bias cannot be assessed in anonymous online surveys, we compared the number of visits (hits) to the study website with the number of completed questionnaires to determine a proxy response rate. Over a 10-week period between May and July 2006, there were 3341 hits to our study website and 1456 completed surveys, resulting in a proxy response rate of 43.58%.

Our final analysis included data from 1291 participants, representing 38.64% of initial survey site hits. A total of 227 participants admitted to lying on the survey (15.59% of respondents). Of these, we excluded 165 from analyses because they lied about their age (n = 127) or reported being less than 18 years of age (n = 38). Of those remaining, 70 reported lying about their weight, so we did not analyze their heights and weights. A total of 18 lied about something other than age and weight, with 13 stating that they had done their best to be truthful but had estimated some answers. The remaining 5 lied on very specific portions of the survey, so we excluded answers to those corresponding questions. We discovered no duplicate responses on analyses. One respondent gave nonsensical responses and was excluded.

With regard to the completeness of responses in crucial data on primary predictor and outcome variables, only 1.63% (n = 21) responses were “break-offs” (surveys with less than 50% of questions answered), and another 3.72% (n = 48) were partially complete (50%-80% of questions answered), according to standard definitions of these values in survey reporting as defined by the American Association for Public Opinion Research [18].

Characteristics of Website Users

Table 1 describes clinical and demographic characteristics reported by participants at the time of the survey, disordered eating behaviors, and views on their own health. Many reported being overweight (282/1188, 23.73%) or obese (406/1188, 34.17%) in the past, with a mean %median BMI at their highest weight of 128.59% (range 70–298, SD 32.71). The mean %median BMI at the lowest reported weight was 88.6% (range 44–233, SD 19.0). While many (800/924, 86.6%) self-diagnosed an eating disorder, 67.62% (873/1291) had never been in treatment, 87.09% (1100/1263) were not currently being treated, and 39.20% (499/1273) had ever had a formal diagnosis of an eating disorder. Underweight participants were more likely to have been treated than normal-weight or overweight respondents (41.6% vs 30.4% or 24.9%, χ22 = 18.9, P < .001).

Table 1. Clinical and demographic characteristics of study participants (N = 1291).
Age (years)1260
Gender (n=1291)



Ethnicity (n=1291)




African American or black302.32

American Indian/Alaskan Native171.32

Native Hawaiian/Pacific Islander40.31


Marital status (n=1285)



Employment/student status (n=1284)



Combined employed/student372.87

Unemployed and not a student12910.00

BMIa (kg/m2)1192
%Median body weight1192
Weight class by CDCbcriteria (n=1192)

Underweight (BMI <18.5)29624.83

Normal weight (BMI 18.5–24.9)64754.28

Overweight (BMI 25–29.9)14812.42

Obese (BMI ≥30)1018.47

EDcdiagnosis if formally diagnosed (n=498)

Anorexia nervosa22545.2

Bulimia nervosa10420.9


Binge eating disorder81.6

Did not specify10.2

Age at dieting onset (years)1236
Age at ED onset (years)1031
Disease duration (years)1006
Activities in the last month

Counting of calories, fat, or carbohydrates (n = 1265)116992.41

Compulsive exercise (n = 1260)96576.59

Secretive eating (n = 1270)95274.96


Ever (n = 1269)96976.36

Last month (n = 1251)71757.31

Age at onset964
Binge eating

Ever (n = 1269)106883.96

Last month (n = 1271)84366.33

Age at onset1019
Laxative use

Ever (n = 1273)70054.99

Last month (n = 1264)44335.05

Age at onset672
Diet pill use

Ever (n = 1271)91772.15

Last month (n = 1265)55143.66

Age at onset909

Used >7 times per week60265.9

Age at first ED treatment (years)402–38.0
Admission to hospitalization

In past 30 days (n = 1243)201.61

Ever (n = 1253)17013.57

Missed menses in last year (n = 1238)




None, but I am on the pill14811.95

I have never had a period120.97

Diagnosis of low bone density (n = 1183)1109.30

Mental health diagnoses

Depression (n = 1271)74858.85

Anxiety (n = 1271)53041.70

ADHDe (n = 1264)13610.76

Other (n = 1121)36732.74

Ever treated with psychiatric medication (n = 1259)64851.47


In last 30 days (n = 1270)45435.75

Ever (n = 1271)98877.73

Cutting (n = 1291)84165.14

Burning (n = 1291)28522.08

Scratching (n = 1291)55843.22

Other (n = 1291)35127.18

Views on own health (n = 1257)

Eating disordered by choice38930.95


Recovering or trying to recover15912.65


Other or combination of above25119.97

Support pro-ED movement (n = 1227)

Completely or very much53943.93

A little bit or somewhat48739.69

Not at all20116.38

a Body mass index (kg/m2).

b US Centers for Disease Control and Prevention.

c Eating disorder.

d Eating disorder not otherwise specified.

e Attention deficit/hyperactivity disorder

Mean scores on EDE-Q subscale and global scores exceeded the 90th percentile for young adult female norms. Mean scores on subscales were 4.56 (SD 1.25, 95–99 percentile) for Restraint, 3.83 (SD 1.29, 95–99 percentile) for Eating Concern, 5.18 (SD 0.94, 90–95 percentile) for Shape Concern, and 4.86 (SD 1.07, 95–99 percentile) for Weight Concern. The EDE-Q global score showed significant pathology with a mean of 4.61 (SD 0.96, 95–99 percentile).

On the CDC Healthy Days Measure, 5.23% (67/1280) indicated their health was poor, 24.14% (309/1280) fair, 39.61% (507/1280) good, 22.42% (287/1280) very good, and 8.59% (110/1280) excellent. Despite these encouraging answers, participants reported a mean of 6.5 (SD 7.8) physically unhealthy days in the past month, 18.4 mentally unhealthy days (SD 9.9), and 21.1 (SD 9.9) unhealthy days overall. On average, participants reported 7.5 (SD 8.3) days in which their activity was limited in the last month. Moreover, 24.00% (248/1033) reported they had to stop school, and 17.49% (192/1098) needed to stop working in the past because of their eating disorder. As well, 38.06% (427/1122) spent less time on their school or work, 36.10% (422/1169) spent less time in recreational activities, and 58.70% (725/1235) spent less time with friends over the past month, owing to their eating disorder.

EDQOL subscales were all in the severely impaired range, with participants demonstrating a mean psychological score of 2.78 (SD 0.74), physical/cognitive score of 1.97 (SD 0.87), financial score of 0.55 (SD 0.82), and work/school score of 0.63 (SD 0.81). Overall, the global quality-of-life score on this measure was 1.49 (SD 0.60), also in the severely impaired range.

The reported average age at onset of visiting pro-ED websites was higher than the reported age at onset of other disordered eating behaviors and years after participants felt their eating disorder had begun (Figure 3).

Figure 3. Self-reported age of onset of dieting behaviors and pro-eating disorder (ED) website usage. Mean age of onset of ED-related activities indicated by center horizontal bar, with vertical standard deviation error bars. Horizontal dashed line represents mean onset of pro-ED usage.
View this figure

Website Usage

When asked from which sources they obtained the most eating disorder information in the last 30 days, 86.65% (1077/1243) of participants reported a website, either pro-ana (611/1243, 49.16%), pro-mia (4.42%, 55/1243), pro-recovery (44/1243, 3.54%), or a general health website (27/1243, 2.17%), and the Internet in general (340/1243, 27.35%). Only a few (39/1243, 3.14%) reported consulting books, and only 1.21% (15/1243) reported consulting physicians. The dominance of the Internet for this population was not surprising, given that 35.73% (453/1268) spent more than 4 hours on the Internet daily, 34.94% (443/1268) spent 2–4 hours each day, and 23.11% (293/1268) spent 1–2 hours each day. Only 6.23% (79/1268) spent less than 1 hour on the Internet daily.

Most participants learned about pro-ED sites by chance (403/1228, 32.82%) or by reading about them (308/1228, 25.08%). A third (466/1269, 36.72%) indicated that they had visited a pro-recovery site in the past 30 days, and 62.20% (790/1270) had ever visited such a site. Table 2 presents additional details regarding the online activities and feelings of this population about pro-ED and pro-recovery websites.

Table 2. Online activities and feelings of the study overall population on pro-eating disorder (pro-ED) websites and of the subgroup using pro-recovery websites.

Pro-ED websitesPro-recovery websites
n%Mean (SD)n%Mean (SD)
Reported activities or feelings

Age first visited (years)1218
18.9 (5.5)735
19.3 (4.8)

Hours/week in last 30 days1147
10.5 (18.2)681
2.3 (6.8)

Have their own site21217.24

Felt supported by website in last 30 daysn = 1210n = 692




A little1159.50

Not at all1058.68

Found a community like self on websiten = 1220n = 705


Very much33527.46


A little15212.46

Not at all1209.84

Most important reason for accessing site in last 30 daysn = 1211n = 658

Motivation for weight loss44736.91

Weight loss tips19215.85

Meeting people635.20

Tips on hiding eating disorder483.96



Help with recovery20.17

Activity at sitesn = 1289n = 790

Read posts112086.89

Visited chat room30723.82

Posted messages69654.00

Read diaries or blogs91370.83

Created my own diary or blog43333.59

Learned about weight loss methods78961.21

Learned about diet pills, laxatives, or weight loss supplements62148.18

Learned about places to purchase new diet pills, laxatives, or weight loss supplements29422.81

Post-website use activity
n = 790

Changed eating habits67154.55

Used new diet pills, laxatives, or weight loss supplements in past 30 days21017.21

Used new diet pills, laxatives, or weight loss supplements ever45437.09

New weight loss or purging methods in past 30 days39132.26

New weight loss or purging methods ever74860.86

Self-harm in past 30 days685.60

Self-harm ever18014.65
Site components that were motivational for continuing eating disorder or recovery

Photographs and artwork109789.70

Forum postings and chat rooms104085.60

Diet and exercise information108689.16

Diaries, journal entries, and blogs101783.50

Website Usage Level and Outcomes

We grouped participants based on the frequency and duration of pro-ED website usage. Light users (n = 199, 16.1%) were those who reported less than 1 month of website usage regardless of frequency, or 1–12 months of usage with a frequency of less than once a month. Heavy users (n = 513, 41.5%) were those who used the websites at least daily and had visited them for 12 or more months. All other participants were considered medium users (n = 525, 42.4%). Mean hours/week of website usage were 3.02 (SD 4.8), 7.8 (SD 12.8), and 16.1 (SD 23.7, P < .001) for light, medium, and heavy users, respectively. Differences between groups in outcomes are presented in Table 3.

Table 3. Pro-eating disorder (pro-ED) website usage and associated eating disorder activities and outcomes.
CharacteristicTotalLight (n = 199)Medium (n = 525)Heavy (n = 513)χ2/FdfP value
Age at survey (years), mean22.023.12221.66.1a2.008
Current %median body weight, mean101.9106.3101.4100.53.5b2.006
Eating Disorder Examination Questionnaire score, mean


Eating concerns3.833.243.714.1743.8c2≤.001

Shape concerns5.184.655.155.449.1c2≤.001

Weight concerns4.864.275.825.1351.6c2≤.001

Healthy days (in last 30 days)

Overall health poor (n = 1280)67 (5.23%)≤.001

Overall health excellent (n = 1280)110 (8.59%)14.196.429.2c8≤.001

Unhealthy physical days, mean6.≤.001

Unhealthy mental days, mean18.416.617.62011.2c2≤.001

Sum unhealthy days, mean21.119.120.422.611.2c2≤.001

Days of limited activities, mean7.577.18.33.0d2.05
Eating Disorder Quality of Life score, mean





Amenorrhea (n = 1082)367 (33.92%)25.128.742.429.2c4≤.001
Low bone mineral density (n = 1183)110 (9.30%)8.98.710.1.5742.750
Age started dieting (years), mean13.314.313.512.616.7c2≤.001
Age eating disorder began (years), mean14.314.714.613.94.2b2.015
Disease duration (years), mean7.
History of eating disorder treatment (n = 1291)419 (32.46%)23.628.039.222.5c2≤.001
History of hospitalization (n = 1253)170 (13.57%)6.612.317.315.1c2.001
Age first binged (years), mean14.514.714.514.5.2382.788
Binge eating

Last 30 days (n = 1271)843 (66.33%)6666.565.8.0512.975

Ever (n = 1272)1068 (83.96%)
Age first purged (years), mean15.916.416.215.64.5b2.012

Last 30 days (n = 1251)717 (57.31%)48.553.265.122.3c2≤.001

Ever (n = 1269)969 (76.36%)66.372.884.432.9c2≤.001
Age first used laxatives (years), mean17.618.417.917.24.3b2.014
Laxative use

Last 30 days (n = 1264)443 (35.05%)28.632.839.89.8a2.007

Ever (n = 1273)700 (54.99%)45.551.562.822.3c2≤.001
Age first used diet pills (years), mean17.118.217.316.78.4c2≤.001
Diet pill use

Last 30 days (n = 1270)551 (43.66%)22.742.153.655.8c2≤.001

Ever (n = 1271)917 (72.15%)53.572.079.748.9c2≤.001
Excessive exercise last 30 days (n = 1260)965 (76.59%)68.474.582.318.0c2≤.001

Last 30 days (n = 1270)454 (35.75%)26.334.041.916.9c2≤.001

Ever (n = 1271)988 (77.73%)66.276.783.023.8c2≤.001
Age first visited a pro-ED site (years), mean18.921.319.517.540.5c2≤.001
Eating-disordered activity at sites

Learned about weight loss methods (n = 1289)789 (61.21%)61.366.561.63.22.200

Learned about diet pills or laxatives (n = 1289)621 (48.18%)35.75253.619.8c2≤.001

Learned about places to purchase new diet pills, laxatives, or weight loss supplements (n = 1289)294 (22.81%)15.122.528.114.1c2.001
Post-website use activity resulting from visiting pro-ED websites

Eating habits have changed (n = 1230)671 (54.55%)40.358.256.119.3c2≤.001

Used new diet pills, laxatives, or weight loss supplements in past 30 days (n = 1220)210 (17.21%)6.217.520.921.8c2≤.001

Used new diet pills, laxatives, or weight loss supplements ever (n = 1224)454 (37.09%)17.934.346.954.0c2≤.001

Used new weight loss or purging methods in past 30 days (n = 1212)391 (32.26%)28.434.631.32.72.253

Used new weight loss or purging methods ever (n = 1229)748 (60.86%)45.16463.423.6c2≤.001

Self-harm in past 30 days (n = 1214)68 (5.60%)

Self-harm ever (n = 1229)180 (14.65%)6.61319.119.6c2≤.001
Host own pro-ED website (n = 1230)212 (17.24%)3.611.927.876.1c2≤.001
Completely support pro-ED (n = 1227)329 (26.81%)7.32536.1135.8c8≤.001

aP < .01; in analysis of variance (ANOVA) testing: post hoc differences between light and medium and light and heavy.

bP < .05; in ANOVA testing: post hoc differences between light and heavy for %median body weight and age when first used laxatives to control weight, between medium and heavy for age when eating disorder began; between light and heavy and medium and heavy for age at first purge.

c P ≤ .001; in ANOVA testing: post hoc differences between all groups for all Eating Disorder Examination Questionnaire scores, Eating Disorder Quality of Life (EDQOL) psychological, physical, and total scores, age started dieting, and age first used pro-ED sites; between light and medium and light and heavy for age first used diet pills; between heavy and light and heavy and medium for EDQOL financial and work/school subscores, and unhealthy physical, mental, and sum days.

Predictors of Disordered Eating and Quality of Life Impairment

To estimate models predicting EDE-Q and EDQOL global scores, we first examined colinearity among the variables of interest. Age and duration of disease were strongly and significantly correlated (r = .76), as were %median body weight and %median BMI at highest (r = .74) and lowest (r = .77) weights. While other variables were correlated, no others were at the point of colinearity. Therefore, we entered age, %median body weight, level of pro-ED website usage, and EDE-Q and EDQOL global scores into our multivariate models.

Significant predictors of EDE-Q scores were EDQOL global score (beta = .56, P < .001) and higher pro-ED usage level (beta = .19, P < .001), predicting 40% of the variance (F4,1116= 185.2, P< .001). Significant predictors of EDQOL scores were EDE-Q scores (beta = .59, P < .001) and age (beta = –.07, P < .005), while pro-ED usage level was not significant (beta = .04, P = .13). This variable set explained 38% of the variance in EDQOL scores (F4,1116 = 167.4, P < .001).

This was the largest and most comprehensive study of adult pro-ED website visitors to date. Most are normal-weight young women, who report multiple extreme weight control behaviors, yet have never been in formal eating disorder treatment. This study challenges presumptions about both pro-ED website users and eating disordered individuals in general.

Our results show a clear association between the level of pro-ED website usage and both disordered eating and quality of life. Pro-ED website usage remains an important predictor of EDE-Q scores even when other commonly reported predictors are considered. Heavy users of pro-ED websites differ significantly from light users; of particular concern are those who spend, on average, around 16 hours per week on these websites. More website usage was strongly associated with higher levels of disordered eating on the EDE-Q and more severe impairment on the EDQOL. Usage level was also incrementally associated with younger age at dieting onset, various disordered eating behaviors, and most harmful post-website use activities, such as diet pill use, weight loss techniques, or self-injury. Nearly a third of heavy website users hosted their own pro-ED website and supported pro-ED as a movement. Website usage was also strongly associated with treatment and hospitalization rates. This was a cross-sectional study, so causality cannot be inferred and may simply reflect the degree of illness in heavy website users. However, the reported average age at onset of visiting pro-ED websites was higher than the reported age at onset of other disordered eating behaviors, and years after participants felt their eating disorder had begun. This suggests a disease progression in which website visitation is a later consequence, and not an earlier cause.

These participants differ from the media’s typical image of underweight teenaged pro-ED users, as over half of participants had been overweight or obese, and a fifth were overweight or obese at the time of the survey. Participants displayed high levels of eating disorder pathology and impaired quality of life on validated measures, consistent with those previously reported for homogeneous populations of more strictly defined anorexia nervosa patients [14,16,18]. Pro-ED website users self-reported dangerous behaviors including purging, laxative use, compulsive exercise, and diet pill use. They also reported high levels of eating disorder sequelae and comorbidities, with nearly 50% reporting menstrual irregularities in the last year. Psychiatric diagnoses were prevalent, with 59% reporting depression, 42% anxiety, 52% a history of treatment with psychiatric medication, and over 75% with a history of self-injury.

Despite the high level of pathology and the majority self-diagnosing an eating disorder, only one-third of participants had ever received formal care for their disordered eating. These findings suggest inadequate screening and diagnosis of eating disorders, and that pro-ED website users are seeking support online instead of with a traditional health model. This is a complex phenomenon, and it has been suggested that pro-ED websites may provide a safe, nonjudgmental, and possibly therapeutic interactive environment [19]. However, they have also been noted to offer advice on managing eating disorders, and may subvert mainstream medical care systems by failing to portray eating disorders as negative conditions requiring professional treatment [20]. Our finding that treatment was much less common in normal-weight, overweight, or obese participants raises the question of whether disordered eating may be more likely to be missed in normal and overweight populations, and that these individuals in particular may seek online support.

Respondents reported an astonishingly high 21.1 unhealthy days in the past month (a normal value is 6 unhealthy days in the past 30 days, with diabetic people averaging 11, and breast cancer patients averaging 12) [15]. Participants also had severe levels of impairment on a disease-specific measure, the EDQOL. These results also indicate the need for further prospective study on quality of life and disordered eating, as they appear to be strongly associated.

Online survey methods are new and developing. This comprehensive study design successfully obtained representation from a wide spectrum of websites meeting specific inclusion criteria based on pro-ED content. Engaging website designers and maintainers to incorporate dedicated links to our study allowed us to include over 1200 participants over a relatively short study period. We also assessed participant truthfulness, which allowed us to further improve the quality of data analyzed. Finally, studying participants via the Internet not only offers a population more diverse than many previously reported clinical samples, but also circumvents the ethical problem of inadvertently introducing potentially harmful website content to a population previously naïve to pro-ED sites.

This study has some important limitations. First, respondents were queried about activities that began many years prior to the study, rendering the possibility of recall bias. Second, this was a convenience sample of users who chose to participate, potentially resulting in a selection bias. Third, all online surveys are limited by an inability to meet with participants to verify responses, although prior studies have noted that truthfulness is actually increased using online survey tools when discussing sensitive topics [21]. Fourth, our survey design did not involve a method of access control and thus could have had duplicate responses from the same IP address. Finally, because so few respondents had received treatment, it is possible that the participants underreported or had not yet identified some medical outcomes.

These findings highlight the need to consider the Internet more often as a vehicle for intervention and study, as it offers easy accessibility to users. If appropriate and helpful interventions are developed, they have the potential to reach many people beyond traditional treatment center walls. Users of pro-ED content websites visit them not just for motivation for weight loss or specific dieting tips, but also for emotional support. Pro-recovery websites do not resonate as well with these users, indicating that the medical community needs to listen to online health seekers to determine whether there are self-help sites or educational modules that this population would find meaningful.

Websites with pro-ED content may play both supportive and harmful roles for those struggling with disordered eating. While the content of these sites may affect the eating behaviors of website visitors, the extent of website usage appears to have a more central role in eating behaviors, weight concerns, and quality of life. Moreover, these findings confirm that many with significantly disordered eating, medical complications, psychiatric comorbidity, and severely impaired quality of life are not accessing traditional care and do not fit conventional eating disorder models. They seek support from a Web-based peer group, which poses both potential harms and opportunities for interventions within these online communities. It is critical that future studies comprehensively address possibilities for intervention and improved relationships with these forums, in order to advance our treatment and screening procedures into an online age.


Access to data: Rebecka Peebles had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. The authors are prepared to provide the data and will cooperate fully in obtaining and providing the data on which the paper is based for examination by the editors or their assignees.

Funding: Rebecka Peebles received funding from the Pediatric Research Fund at Stanford University School of Medicine. Jenny Wilson received funding from the Medical Scholars Research Program through Stanford University School of Medicine for her research. James Lock received funding from the National Institutes of Health K24 mechanism. No sponsor had any role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.

Previous presentation: The authors certify that the manuscript represents valid work and that neither this manuscript nor one with substantially similar content under their authorship has been published or is being considered for publication elsewhere.

Other contributors: The authors wish to thank all participants and website maintainers who participated in our study. We also wish to thank all research assistants at the Stanford WEIGHT Lab for their assistance in data collection and coding. Finally, we wish to thank Scott Engel, PhD, at the Department of Psychology, North Dakota State University, Fargo, ND, USA, for use of the EDQOL, and Christopher Fairburn, DM, Department of Psychiatry, University of Oxford, England, for use of the EDE-Q.

Authors' Contributions

Rebecka Peebles was involved in all components of this study including conception, design, data acquisition and interpretation, drafting, revising and approval of the manuscript, and the integrity of the work as a whole. Jenny Wilson was involved in the conception, design, data acquisition and interpretation, and drafting, revising, and approval of the manuscript. Dina Borzekowski was involved in the conception and design, analysis and interpretation of data, drafting and critically revising the manuscript, supervision, and approval of the published version. Kristina Hardy was involved in conception and design, analysis and interpretation of data, manuscript revision, and approval of the final version. James Lock was involved in the project conception and design, analysis and interpretation of data, critical revision, supervision, and approval of the final version. Julia Mann was involved in the conception and design, acquisition of data, analysis and interpretation of data, drafting the manuscript, and approval of the final version. Iris Litt was involved in the study conception and design, interpretation and analysis of data, critical revision of the manuscript, supervision, and approval of the final version.

Conflicts of Interest

None declared.

Multimedia Appendix 1

Online survey. Note: The survey was one continuous online page.

PDF File (Adobe PDF File), 103KB

  1. Wilson JL, Peebles R, Hardy KK, Litt IF. Surfing for thinness: a pilot study of pro-eating disorder Web site usage in adolescents with eating disorders. Pediatrics 2006 Dec;118(6):e1635-e1643 [FREE Full text] [CrossRef] [Medline]
  2. Borzekowski DL, Schenk S, Wilson JL, Peebles R. e-Ana and e-Mia: A content analysis of pro-eating disorder Web sites. Am J Public Health 2010 Aug;100(8):1526-1534. [CrossRef] [Medline]
  3. Harshbarger JL, Ahlers-Schmidt CR, Mayans L, Mayans D, Hawkins JH. Pro-anorexia websites: what a clinician should know. Int J Eat Disord 2009 May;42(4):367-370. [CrossRef] [Medline]
  4. Borzekowski DL, Bayer AM. Body image and media use among adolescents. Adolesc Med Clin 2005 Jun;16(2):289-313. [CrossRef] [Medline]
  5. Norris ML, Boydell KM, Pinhas L, Katzman DK. Ana and the Internet: a review of pro-anorexia websites. Int J Eat Disord 2006 Sep;39(6):443-447. [CrossRef] [Medline]
  6. Abbate Daga G, Gramaglia C, Pierò A, Fassino S. Eating disorders and the Internet: cure and curse. Eat Weight Disord 2006 Jun;11(2):e68-e71. [Medline]
  7. Andrist LC. Media images, body dissatisfaction, and disordered eating in adolescent women. MCN Am J Matern Child Nurs 2003;28(2):119-123. [Medline]
  8. Csipke E, Horne O. Pro-eating disorder websites: users' opinions. Eur Eat Disord Rev 2007 May;15(3):196-206. [CrossRef] [Medline]
  9. Custers K, Van den Bulck J. Viewership of pro-anorexia websites in seventh, ninth and eleventh graders. Eur Eat Disord Rev 2009 May;17(3):214-219. [CrossRef] [Medline]
  10. Jett S, LaPorte DJ, Wanchisn J. Impact of exposure to pro-eating disorder websites on eating behaviour in college women. Eur Eat Disord Rev 2010;18(5):410-416. [CrossRef] [Medline]
  11. Bardone-Cone AM, Cass KM. What does viewing a pro-anorexia website do? An experimental examination of website exposure and moderating effects. Int J Eat Disord 2007 Sep;40(6):537-548. [CrossRef] [Medline]
  12. Harper K, Sperry S, Thompson JK. Viewership of pro-eating disorder websites: association with body image and eating disturbances. Int J Eat Disord 2008 Jan;41(1):92-95. [CrossRef] [Medline]
  13. MacManus R. ReadWriteWeb. 2006 Jan 9. Latest SNS Numbers: MySpace Streaks Ahead   URL: [accessed 2012-08-23] [WebCite Cache]
  14. Fairburn CG, Beglin SJ. Assessment of eating disorders: interview or self-report questionnaire? Int J Eat Disord 1994 Dec;16(4):363-370. [Medline]
  15. Centers for Disease Control and Prevention. 2011 Mar 15. Health-Related Quality of Life (HRQOL): Key Findings   URL: [accessed 2011-12-08] [WebCite Cache]
  16. Engel SG, Wittrock DA, Crosby RD, Wonderlich SA, Mitchell JE, Kolotkin RL. Development and psychometric validation of an eating disorder-specific health-related quality of life instrument. Int J Eat Disord 2006 Jan;39(1):62-71. [CrossRef] [Medline]
  17. Centers for Disease Control and Prevention. 2010 Sep 09. Growth Charts   URL: [accessed 2011-12-08] [WebCite Cache]
  18. Mond JM, Hay PJ, Rodgers B, Owen C, Beumont PJ. Validity of the Eating Disorder Examination Questionnaire (EDE-Q) in screening for eating disorders in community samples. Behav Res Ther 2004 May;42(5):551-567. [CrossRef] [Medline]
  19. Dias K. The Ana Sanctuary: women's pro-anorexia narratives in cyberspace. J Int Womens Stud 2003 Apr;4(2):31-45 [FREE Full text]
  20. Fox N, Ward K, O'Rourke A. Pro-anorexia, weight-loss drugs and the internet: an "anti-recovery" explanatory model of anorexia. Sociol Health Illn 2005 Nov;27(7):944-971. [CrossRef] [Medline]
  21. Kam LE, Chismar WG. Online self-disclosure: model for the use of internet-based technologies in collecting sensitive health information. Int J Healthc Technol Manag 2006;7(3-4):218-232. [CrossRef]

BMI: body mass index
CDC: Centers for Disease Control and Prevention
EDE-Q: Eating Disorder Examination Questionnaire
EDQOL: Eating Disorder Quality of Life
pro-ED: pro-eating disorder

Edited by G Eysenbach; submitted 16.06.12; peer-reviewed by L Fernandez-Luque; comments to author 11.07.12; revised version received 13.07.12; accepted 13.07.12; published 25.10.12


©Rebecka Peebles, Jenny L Wilson, Iris F Litt, Kristina K Hardy, James D Lock, Julia R Mann, Dina LG Borzekowski. Originally published in the Journal of Medical Internet Research (, 25.10.2012.

This is an open-access article distributed under the terms of the Creative Commons Attribution License (, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on, as well as this copyright and license information must be included.