Skip to main content
NCBI home page
VA Author Manuscripts logoLink to VA Author Manuscripts
. Author manuscript; available in PMC: 2026 Feb 4.
Published in final edited form as: Eat Behav. 2024 Jan 23;52:101846. doi: 10.1016/j.eatbeh.2024.101846

Validation of the EDDS-5 self-report survey against the SCID-5 diagnostic interview in US veterans☆

Shira Maguen a,b,*, Adam Batten a,b, Sarah E Siegel a,b, Joy Huggins a,b, Jennifer L Snow c, Lindsay M Fenn c, Alexandra M Dick a, Christiane Zenteno a, Anna C West a, Robin M Masheb c,d
PMCID: PMC12866619  NIHMSID: NIHMS2124340  PMID: 38335645

Abstract

The aim of our study was to validate the Eating Disorder Diagnostic Scale (EDDS-5) updated for the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) with a diverse veteran population against a clinician-administered interview based on the Structured Clinical Interview for DSM-5 (SCID-5). Our sample included 343 veterans, 18–75 years, recruited April 2019 to December 2022 who completed the EDDS-5 as well as other eating disorder and mental health measures. A subsample of these veterans received clinical interviews (n = 166), which were used to validate the EDDS-5. We found that despite multiple proposed modifications, the EDDS-5 performed poorly at correctly identifying diverse veterans who were diagnosed as having eating disorders through clinician-administered interviews. The sensitivity was very low, indicating that using the EDDS-5 did not identify many true positives and may also over diagnose those without true eating disorders. The EDDS-5 may not be the best for screening or diagnostic purposes among diverse samples like veterans.

Keywords: Eating disorders, Diagnostic, Veteran, Clinical interview, Diversity

Eating disorders are a major public health burden in the United States (US). At least 30 million people have an eating disorder in the US (Hudson et al., 2007). While eating disorder research has primarily focused on Anorexia Nervosa (AN) and Bulimia Nervosa (BN), much of this prior work focused on White, adolescent girls and college age women, limiting our understanding of the prevalence and reach of eating disorders. The Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) has expanded to include eating disorders such as Binge Eating Disorder (BED) and multiple Other Specified Feeding and Eating Disorders (OSFED) that may better capture eating disorders for more diverse populations. OSFED includes Night Eating Syndrome (NES), Atypical Anorexia (AA), and BN or BED of low frequency and/or limited duration.

The revision and addition of DSM-5 diagnoses allows for greater identification of eating disorders in more diverse groups, including veterans (Masheb et al., 2021). BED is widely prevalent among veterans (Higgins et al., 2013), NES affects 10 % of overweight veterans (Dorflinger, Ruser, & Masheb, 2017), and AA was observed in 18 % of female post-9/11 veterans (Masheb et al., 2021). Binge eating is prevalent across racial and ethnic groups (Franko et al., 2012; Thompson-Brenner et al., 2013), disproportionately affects male veterans, compared to female veterans (Higgins et al., 2013), and is present in as many as 78 % of overweight or obese veterans seeking care through a national weight management program (Higgins et al., 2013).

A honed assessment tool is critical for healthcare systems to identify individuals with eating disorders. The Eating Disorder Diagnostic Scale updated for the DSM-5 (EDDS-5; Stice, Telch, & Rizvi, 2000), is a 23-item self-report scale assessing AN, BN, BED, and OSFED through questions about eating behaviors, appearance-related cognitions, and compensatory behaviors over a three-month period. Scoring for the EDDS-5 allows for self-reported, probable diagnosis of each of these eating disorders. The EDDS-5 assesses eating disorders without clinician administration, thereby saving time and costs associated with lengthier clinical interviews. While not replacing the importance of a clinician diagnosis, tools like the EDDS-5 can help identify patients who potentially need specialist care.

The EDDS-5 has been used in multiple studies, including those with veterans (Masheb et al., 2021; Mitchell et al., 2021) to identify probable eating disorders. However, the EDDS-5 has never been validated in any sample. The prior version of the EDDS has been validated with different groups (e.g., bariatric patients, Hong Kong adolescents, etc.; Kelly et al., 2012; Schaefer et al., 2019; Stice et al., 2000; Stice, Fisher, & Martinez, 2004; Williams et al., 2017), but has not been validated with veterans, a more diverse group than the adolescents and college-age women predominant in prior validation studies (Stice et al., 2000, 2004). One study that included Black and White women and used confirmatory factor analyses found that the EDDS factor loadings were not invariant across these groups, concluding that the EDDS does not assess equivalent constructs in White and Black women (Kelly et al., 2012). Given that men were not part of the original validation, another study included college age men when using the EDDS and found that additional items were needed to reflect eating disorder symptoms that more commonly occur in men (e.g., muscularity-oriented disordered eating; Schaefer et al., 2019). Consequently, there is evidence that the EDDS may not perform as well with race and gender diverse samples.

Given recent evidence that eating disorder prevalence is higher than previously estimated in diverse groups (Dorflinger et al., 2017; Franko et al., 2012; Masheb et al., 2021; Thompson-Brenner et al., 2013), validating the EDDS-5 with a high-risk eating disorder population is an important first step to ensuring its generalizability. Our goal was to validate the EDDS-5 with a diverse group of veterans. Veteran samples are diverse in age, race/ethnicity, gender, and eating disorder presentation. This is the first study ever to validate the EDDS-5 in any population and validate it against a clinician-administered interview based on the Structured Clinical Interview for DSM-5 (SCID). Our aim was to analyze the concordance between the EDDS-5 self-administered questionnaire and SCID interview and analyze the concordance between alternative diagnostic scoring methods of the EDDS-5 and SCID.

1. Methods

1.1. Participants

We surveyed veterans seen in the San Francisco (SF) and Connecticut (CT) VA Healthcare Systems (HCS) between 18 and 75 years of age from April 2019 to December 2022. Exclusion criteria included current psychosis-related diagnoses in the VA electronic medical records (Corporate Data Warehouse (CDW), Fihn et al., 2014). We used chart review and an initial screening questionnaire to identify additional exclusion criteria, including those actively suicidal within the last year (those with suicidal ideation were not excluded, only those with acute intent or a plan), pregnant or breastfeeding, experiencing unstable housing, or taking medication intended to affect appetite, eating, or weight. Veterans who met inclusion criteria were mailed or emailed recruitment letters with the option to participate or opt-out. If veterans indicated interest, they were called, participation was explained, and verbal consent was obtained. Those who did not respond to the recruitment letter were called after a two-week waiting period. If they responded to outreach, the study was explained, and they were invited to participate. The study survey was either mailed with a return envelope or emailed to participants with a link to the online version. Participants were compensated $25 upon completion and return of the survey. All procedures were approved by the University of California, San Francisco Institutional Review Board, the Human Research Protection Program at the San Francisco VA Health Care System, and the Institutional Review Board at the VA Connecticut Healthcare System.

Of the 5826 invitations sent, 489 (8 %) responded and 343 completed >75 % of the ED-specific items in the Qualtrics survey for a completer response rate of 6 % (see Fig. 1). We hypothesize that survey fatigue documented during the pandemic may have impacted study responses (de Koning et al., 2021). Semi-structured, clinician-administered interviews were conducted until a validation sample consisting of 78 individuals diagnosed with an eating disorder and a stratified randomly selected control group of participants who did not have an eating disorder were completed. This number of interviews was stipulated in our power and sample calculations. To increase our chances of reaching the 78 diagnosed eating disorder participants as quickly as possible, survey participants were chosen for interview if they met a high likelihood of eating disorders (HLED) profile described below.

Fig. 1.

Fig. 1.

Open in a new tab

Participant flow chart for survey and validation phases.

Note: SFVA = San Francisco VA; VACT = VA Connecticut; HLED = high likelihood of eating disorders.

To determine the number of participants to interview, we assumed the same prevalence of ED+ (20 %), sensitivity (91 %), and specificity (53 %) from a prior review, and the same ±10 % width of a 2-sided 95 % CI but using a Type I error of 0.10. We estimated that 110 interviewees would be sufficient for our analysis (Bartlett & Mitchell, 2015; Buderer, 1996).

1.2. Measures

The survey included sociodemographic variables, weight history, and multiple mental health and eating disorder survey measures, including the EDDS-5. We administered both self-report measures and clinician-administered interviews (see Table 1). We applied extensive cleaning and quality checks on all items prior to scoring and analysis.

Table 1.

Study self-report measures and clinician-administered assessments.

Variable Assessment
Sociodemographic Characteristics Demographics: Age; gender; biological sex; race/ethnicity; sexual orientation; marriage status; highest level of education; employment status; disability status; total household income.
Military History: Service era; service branch; component type; rank; job title; combat specialty; total years served; times deployed to a war zone; year of separation from service; age and weight at separation from service; weight one year after separation from service.
Validation Measure EDDS-5: The Eating Disorder Diagnostic Scale (Stice et al., 2000) is an 23-item self-report measure that assesses a time frame of past three months for DSM-5 criteria of anorexia nervosa, bulimia nervosa, binge eating disorder, purging disorder, night eating syndrome, atypical anorexia nervosa, and low frequency bulimia nervosa and binge eating disorder. Ratings are on a seven-point Likert scale, ranging from 0 (not at all) to 6 (extremely).
Weight History Weight History: Height; weight; highest & lowest adult weight; first overweight.
Self-Recognition: Past & present ED diagnosis; ideal weight.
Treatment History: Past & present treatment status.
Eating Behaviors and Cognitions MOVE-11: Endorsement of one time a week or more to a single item, “On average, how often have you eaten extremely large amounts of food at one time and felt that your eating was out of control at that time?” was considered a positive screen (Dorflinger et al., 2017).
EDE-Q: The Eating Disorder Examination-Questionnaire (Fairburn & Beglin, 1994) is a 37-item self-report measure that assesses a time frame of past-month for cognitions and behaviors related to bulimia nervosa and anorexia nervosa in addition to some binge eating disorder symptoms. Ratings are on a seven-point Likert scale ranging from 0 (no days or not at all) to 6 (every day or markedly) with six items that specify number of days in the last month.
NEQ: The Night Eating Questionnaire (Allison et al., 2008) is a 14-item self-report measure that assesses night eating syndrome symptoms. Ratings are on a five-point Likert scale, each item with different response options that correspond to the question. Stop criteria are built into the measure to reduce burden on participants who indicate no nighttime awakenings or snacking in the night.
Diagnostic Interview (n = 166) SCID-5-RV: The Structured Clinical Interview for DSM-5 (First, Williams, Karg, & Spitzer, 2015) is a standardized diagnostic interview to assess psychiatric disorders based on DSM-5 criteria. For this study, we used the following modules: depression, bipolar, psychosis, substance use, anxiety, trauma, obsessive- compulsive & related, sleep-wake, and feeding & eating.
Supplement 1: EDE: The Eating Disorder Examination (Cooper & Fairburn, 1987) is a semi-structured interview that assesses a time frame of past three months for symptoms of bulimia nervosa and anorexia nervosa with additional questions related binge eating disorder symptoms. There are four subscales: Restraint (intentions and attempts to control/restrict food intake), Eating Concern (preoccupation with eating, fear of loss of control over eating), Weight Concern (attitudes around weight loss desire, avoidance of weight gain), and Shape Concern (pursuit of thin shape, body dissatisfaction). The interview provides an overall global score by averaging the four subscale scores.
Supplement 2: NESHI: The Night Eating Syndrome History and Inventory (Lundgren et al., 2012) is a semi-structured interview based on the NEQ that assesses the presence and frequency of night eating symptoms more comprehensively than the survey alone and allows for diagnosis of NES to be made.
Bias Adjustment Instruments EDE-Q: See above
EAT-26: The Eating Attitudes Test (Garner, Olmsted, Bohr, & Garfinkel, 1982) is a widely used, 26-item self-report measure that assesses general eating disorder pathology. Ratings are on a six-point Likert scale, with values of 0 (sometimes, rarely, & never), 1 (often), 2 (usually), and 3 (always), with several items reverse- scored.
WBIS-M: The Modified Weight Bias Internalization Scale (Pearl & Puhl, 2014) is an 11-item self-report measure of the degree to which a respondent believes negative stereotypes and statements about overweight/obese people apply to themselves. Ratings are on a seven-point Likert scale, ranging from 1 (strongly disagree) to 7 (strongly agree). One item is reverse scored.
LOCES-7: The Loss of Control over Eating Scale ( Latner, Mond, Kelly, Haynes, & Hay, 2014) is a 7-item self-report measure that assesses aspects of LOC-eating in clinical and non-clinical populations. Ratings are on a five-point Likert scale, ranging from 1 (never) to 5 (always).
M-YFAS: The Modified Yale Food Addiction Scale ( Gearhardt, Corbin, & Brownell, 2009) is a 13-item self-report measure of addictive-like eating behavior. Ratings are on an eight-point Likert scale ranging from 0 (never) to 2 (once a month), 4 (once a week), and 7 (every day).
DEQ-6: The Discreet Emotions Questionnaire 6-item ( Rudich, Lerman, Gurevich, Weksler, & Shahar, 2008) is a self-report measure of self-criticism. Ratings are on a seven-point Likert scale, ranging from 1 (strongly disagree) to 7 (strongly agree).
PHQ: The Patient Health Questionnaire (Spitzer, Kroenke, & Williams, 1999) is a commonly used screening measure for eight mental health disorders: major depressive disorder, panic disorder, other anxiety disorder, other depressive disorder, probable alcohol abuse or dependence, somatoform, as well as eating disorders.
PCL-5: The PTSD Checklist for DSM-5 (Weathers et al., 2013) is a 20-item self-report measure of DSM-5 symptoms of PTSD. Ratings are on a five-point Likert scale, ranging from 0 (not at all) to 4 (extremely). The version used here included the Criterion A component wherein participants reported the event that initiated their symptoms.
VR-12: The Veterans RAND 12-item Health Survey ( Selim et al., 2009, pg.12) is a self-report measure assessing quality of life (both mental health and physical health components).
ISI: The Insomnia Severity Index (Bastien, Vallières, & Morin, 2001) is a 7-item self-report measure that assesses the nature, severity, and impact of insomnia. Ratings are on a five-point Likert scale, each item with different response options that correspond to the question.
DRRI: The Deployment Risk and Resilience Inventory-2 (Vogt et al., 2013) is a 15-item self-report measure of combat exposure and military experiences. Responses are dichotomous yes/no.
Open in a new tab

1.2.1. Validation sample

To increase the likelihood of selecting candidates who would meet diagnostic criteria for a DSM-5 eating disorder, we sequentially selected at-risk and not at-risk survey participants to interview from our main survey sample until we reached our validation sample size for each risk group as determined by our power and sample calculations detailed above. At-risk participants were considered for an interview if they had a high likelihood of eating disorders (HLED) as indicated by endorsement of one or more of the following indicators: self-evaluation of weight or shape; weight suppression; compensatory behaviors; binge eating; or night eating (see supplement for specific items). Other than this procedure, the inclusion/exclusion criteria were the same as our overall sample.

1.3. DSM-5 structured interview (enhanced SCID)

To achieve the most comprehensive diagnostic assessment for eating disorders we utilized the Eating Disorder module of the SCID Interview and enhanced this with the EDE and NESHI semi-structured diagnostic interviews for eating disorder and night eating diagnoses, respectively (see Table 1; modified assessments available from authors upon request).

More specifically, SCID prompts were reviewed and enhanced to ensure they would capture DSM-5 eating disorder diagnoses for a military and veteran population. The following changes were made to enhance assessments of self-evaluation, binge episodes, diagnoses of low frequency and/or limited duration, and NES. The SCID DSM-5 criterion for self-evaluation unduly influenced by body shape or weight asks, “At your lowest weight, did you still feel too fat or that part of your body was too fat?” The research team agreed this query did not address self-evaluation that is tied to self-worth, most relevant for this population and as indicated in DSM-5. Thus, we added the EDE query, “Has your shape or weight influenced how you think about (judge) yourself as a person?” rated on a scale from 0 to 6 (“not at all” to “markedly”) with scores of 4 or higher considered positive for this criterion.

SCID assessment of binge eating was enhanced given our clinical experiences with men and those with military backgrounds suggesting discomfort and avoidance of acknowledging feelings of being out of control. The following prompts were made available to interviewers to assess “out of control: ” Are you resigned to have this episode of overeating? Are these episodes inevitable/going to happen anyway? Are these episodes on auto-pilot? Do you give up trying to control the episodes? To provide SCID assessors with more detailed guidance and participant data on the size of binge episodes, the EDE Appendix for size of binge episodes was used, and if needed, the EDE instructions for reviewing a day of eating were followed. Finally, BED and BN diagnoses of low frequency or limited duration were refined such that the binge eating (and compensatory behaviors) were considered low frequency if they occurred less than once a week but at least once a month, and low duration if they occurred for less than three months but at least one month.

Given the absence of SCID prompts to assess NES, an abridged NESHI interview was developed by one of the investigators (RM) with the NESHI creator (Kelly Allison). Abridged NESHI queries were converted from the original response sets to dichotomous responses for the presence of the following criteria over the past 7 days: 1) 25 % of food intake was consumed after the evening meal, or eating occurred during the night after having been asleep (yes/no for each), 2) awareness and recall of evening and nocturnal eating episodes (none/somewhat or more), and 3) presence of at least three of five features (yes/no for each; lack of desire for breakfast/to eat in the morning at least 4× per week, strong urge to eat between dinner and sleep onset and/or during the night, sleep onset and/or sleep maintenance insomnia are present four or more nights per week, presence of a belief that one must eat in order to initiate or return to sleep, and mood is frequently depressed and/or mood worsens in the evening). In addition, the abridged NESHI assessed association with significant distress and/or impairment in functioning (yes/no for each), presence for at least a three-month duration (dichotomized from years and months symptoms were present), and not due to substance abuse or dependence, medical disorder, medication, or another psychiatric disorder (did the night eating start after one of the aforementioned problems; yes/no for each).

Evaluators were blind to the HLED classification algorithm, trained in SCID-enhanced administration and modifications, and received ongoing supervision. Diagnoses included: AN, BN, BED and OSFED. OSFED diagnoses include (1) Atypical Anorexia Nervosa: All AN criteria are met, but weight is at or above the normal range. (2) BED of low frequency and/or limited duration: All BED criteria are met, but at a lower frequency and/or for <3 months. (3) BN of low frequency and/or limited duration: All BN criteria are met, but binge eating and compensatory behavior occur at a lower frequency and/or < 3 months. (4) Purging Disorder: Ongoing purging behavior without binge eating; and (5) Night Eating Syndrome: Ongoing episodes of eating after awakening from sleep, or majority of caloric intake occurs in the evening/night. Both Lifetime and Current diagnoses were ascertained. SCID-enhanced notes and diagnoses of each participant were reviewed by one of the authors (RM) and criterion in need of elaboration or confirmation was discussed with the evaluator. In cases requiring further clarity, methods from the EDE were used such as obtaining a full day of eating and making determinations for size and loss of control of binge eating episodes. Given that night eating symptoms are not detailed in the SCID, a modified NESHI (Allison et al., 2008) was created with the developer and used to obtain NES diagnoses.

1.4. Validation

We conducted two validation analyses using diagnoses from the SCID-enhanced interview. In the first analysis, we included anyone with a lifetime eating disorder, including current (Any-SCID). In the second analysis we focused on respondents with a current eating disorder (Current-SCID; last month).

1.5. Diagnostic variables

The EDDS-5 was scored as originally conceived (Stice et al., 2000) and with a number of iterative modifications. For each scoring method, the nine probable eating disorder (PED) diagnoses were collapsed into dichotomous variables (positive if any one of the nine was captured on the EDDS-5 and negative if none of the nine PED diagnoses were captured). The diagnosis from each EDDS-5 modification (EDDS5-PED1–EDDS5-PED3) was then compared to the diagnosis from the SCID interview (SCID-ED) to see which had the highest concordance.

1.6. Data analysis

1.6.1. Demographic and military service

We conducted one-way tests (Chi-squared and t-tests) for all demographic and military service items, and select constructed variables across strata defined by healthcare system (Agresti, 2012; Yoshida, Bohn, & Yoshida, 2020). In all one-way tests, we consider a p-value <0.05 to denote a significant difference across strata.

1.6.2. EDDS-5 reliability

We assessed item correlation (ρ) and two measures of inter-item consistency: Guttman’s lambda (λ6) and Cronbach’s alpha (α) to determine the reliability of the EDDS-5 measure (Cronbach, 1951; Guttman, 1945; W. Revelle & Condon, 2019; W. R. Revelle, 2017).

1.6.3. Concordance

In general, whenever a study is designed to validate the health status of a patient based on the results of a diagnostic test, but only a sample of the patients are selected for validation there is a risk that higher proportions of symptomatic patients will be selected (Begg & Greenes, 1983). This validation bias can potentially result in an inflated sensitivity reported for the diagnostic test. In the current case, the PED variable, and the sampling process itself led to the intentional oversampling of ED+ respondents.

To attenuate the potential risk of bias, we created weights based on the probability of responding to the SCID-enhanced interview using a covariate balancing propensity score (CBPS) method in an inverse probability weighting framework (Austin & Stuart, 2015; Imai & Ratkovic, 2014). In the first step, select items in the survey were run through a Principal Components Analysis (PCA) to identify items that accounted for large amounts of variation in the data. These items were then used as predictor variables in a CBPS model of the probability of responding to the SCID request while balancing the important items discovered in step 1 across the response and non-response groups. In the final step, the predicted probabilities from the CBPS are inverted to create the Inverse Probability of Response Weights (IPRW). Covariates used in calculating the weights included: age, sex, race, ethnicity, years of military service, number and months of war zone deployments, time since leaving the military (years), BMI, whether the respondent completed a paper survey, the recruitment list and sampling wave, HLED algorithm, and several total scores from the mental and physical health and ED domains in the survey (see Table 1 for measures). These covariates include demographic, military, clinical, and procedure variables that prior studies have found to be important to examine as covariates with these outcomes or procedural variables that we wanted to ensure were attenuating any biases that could occur due to wave, online vs. paper surveys, etc.

1.6.4. EDDS-5 validation

Using the inverse probability of response (IPR) weights described above, we compared the three EDDS5-PED modifications to the SCID-ED outcomes using weighted classification tables (Lumley, 2011). From these tables, we calculated sensitivity, specificity, and the concordance index (often referred to as the Area Under the Receiver Operating Curve or AUC) for each variable (Agresti, 2012). To better align with our one-way tests and sample design, we report means and 90 % confidence intervals. All analysis was conducted in R Version 4.0: a language and environment for statistical computing (R Core Team, 2020).

2. Results

2.1. Participant characteristics

Respondents were nearly balanced across the two sites with 185 (53.8 %) and 159 (46.2 %) in the CT and SF HCSs, respectively (Table 2). There were significant differences in sexual orientation (p-value = 0.005) and college education (p-value = 0.037), across sites. In both cases, SF had higher proportions. Additionally, respondents were balanced across interview status (Supplemental Table S1). For continuous measures, CT respondents had significantly higher: average current weight of 196 lbs. vs 183 lbs. in SF (p-value = 0.008); average highest weight of 223 lbs. vs 203 lbs. in SF (p-value = 0.001); and average BMI with 31 in CT vs 29 in SF (p-value = 0.014). These differences in means and proportions may be due to the higher number of veteran men in the CT sample (53 % in CT vs 44 % in SF; p-value = 0.132). Finally, the 166 patients interviewed were balanced across HCS: CT (n = 87, 47 %) and SF (n = 79, 50 %) p-value = 0.70.

Table 2.

Respondent characteristics and one-way tests for proportions and/or means

Site
Overall CT SF p Missing
n 344 185 159
Age (mean (SD)) 56.8 (13.28) 57.5 (13.15) 56 (13.42) 0.281 0.6
Age at separation: (mean (SD)) 29.36 (12.7) 29.72 (15.5) 28.95 (8.47) 0.586 3.8
Male (%) 166 (48.7) 97 (52.7) 69 (43.9) 0.132 0.9
Race (%) 0.064 0
 Black of African 44 27 17
 American (12.8) (14.6) (10.7)
 More than one race 21 (6.1) 10 (5.4) 11 (6.9)
 Other 19 (5.5) 5 (2.7) 14 (8.8)
 White 260 (75.6) 143 (77.3) 117 (73.6)
Ethnicity = not Hispanic (%) 301 (90.9) 164 (93.7) 137 (87.8) 0.094 3.8
Straight or heterosexual (%) 290 (84.3) 166 (89.7) 124 (78.0) 0.005 0
Married currently (%) 165 (48.2) 94 (51.4) 71 (44.7) 0.258 0.6
Education college (%) 181 (52.8) 87 (47.3) 94 (59.1) 0.037 0.3
Employment (%) 0.689 0
 Employed full-time 121 (35.2) 65 (35.1) 56 (35.2)
 Employed part-time 25 (7.3) 10 (5.4) 15 (9.4)
 Retired 107 (31.1) 59 (31.9) 48 (30.2)
 Unemployed or other employment 47 (13.7) 27 (14.6) 20 (12.6)
 VA Service-connection disability 44 (12.8) 24 (13.0) 20 (12.6)
If you are unemployed, do you have a health-related disability that makes it difficult or impossible for you to work? (%) 0.06 2.6
Not applicable 218 (65.1) 117 (65.0) 101 (65.2)
No 62 (18.5) 27 (15.0) 35 (22.6)
Yes 55 (16.4) 36 (20.0) 19 (12.3)
Income = 0-$59,999 (%) 158 (48.6) 89 (50.3) 69 (46.6) 0.585 5.5
Military Service Period Vietnam (%) 109 (31.7) 62 (33.5) 47 (29.6) 0.503 0
Military Service Period 1975 To 1990 (%) 117 (34.0) 58 (31.4) 59 (37.1) 0.313 0
Military Service Period Persian Gulf (%) 71 (20.6) 33 (17.8) 38 (23.9) 0.211 0
Military Service Period 1991 To 2001 (%) 82 (23.8) 36 (19.5) 46 (28.9) 0.054 0
Military Service Period OEF/OIF/OND (%) 102 (29.7) 51 (27.6) 51 (32.1) 0.427 0
Military Service Period Other (%) 72 (20.9) 37 (20.0) 35 (22.0) 0.746 0
Military Service Branch(%) 0.908 4.7
 Air Force 72 (22.0) 38 (21.1) 34 (23.0)
 Army 151 (46.0) 85 (47.2) 66 (44.6)
 Marine Corps 28 (8.5) 14 (7.8) 14 (9.5)
 Navy 77 (23.5) 43 (23.9) 34 (23.0)
Reserves or National Guard (%) 39 (11.4) 24 (13.0) 15 (9.5) 0.4 0.3
Enlisted (%) 300 (87.2) 167 (90.3) 133 (83.6) 0.095 0
How many years total have you served in the Military? (Mean (SD)) 8.84 (8.34) 8.89 (8.63) 8.79 (8.04) 0.916 5.2
How many times have you been deployed to a war zone? (Mean (SD)) 0.79 (3.08) 0.67 (1.14) 0.95 (4.39) 0.406 2.6
How many months combined were you deployed to a war zone? (Mean (SD)) 4.58 (7.76) 4.98 (8.16) 4.13 (7.28) 0.325 4.4
What year did you separate from military service? (Mean (SD)) 1985 (16.17) 1990 (17.1) 1980 (15.75) 0.383 4.9
Current weight (mean (SD)) 189.9 (46.5) 195.9 (49.7) 182.7 (41.4) 0.008 0.3
Current height (mean (SD)) 66.81 (4.14) 66.87 (4.27) 66.75 (4.00) 0.799 1.2
BMI (mean (SD)) 29.79 (6.77) 30.63 (7.02) 28.83 (6.35) 0.014 1.5
Lowest weight (mean (SD)) 142.1 (27.5) 144.2 (29.7) 139.7 (24.5) 0.134 0.6
Highest weight (mean (SD)) 214.2 (54.6) 223.4 (58.5) 203.5 (47.6) 0.001 0.3
Item-HLED = 1 (%) 133 (38.9) 75 (41.0) 58 (36.5) 0.459 0.6
Completed SCID Interview (%) 166 (48.3) 87 (47.0) 79 (49.7) 0.701 0
Current SCID-ED = 1 (%) 74 (44.6) 44 (50.6) 30 (38.0) 0.14 51.7
Any SCID-ED = 1 (%) 99 (59.6) 57 (65.5) 42 (53.2) 0.144 51.7
Open in a new tab

Note: CT = Connecticut; SF = San Francisco. HLED = high likelihood of eating disorders; BMI = body mass index; SCID-ED = Structured Clinical Interview for DSM-5, Eating disorders module; SD = standard deviation.

2.2. Reliability

The EDDS-5 items were all moderately correlated and in the same direction (Supplemental Table S2 and Fig. S1). They also exhibited high inter-item consistency as measured by Cronbach’s α and Guttman’s λ6. (Supplemental Table S2).

2.3. Covariate balance

The IPR model had good fit (C = 0.86) and all variables used in the IPR were balanced after weighting with all standardized mean differences between −0.2 and 0.2 (Supplemental Fig. S2).

2.4. EDDS-5 modifications

2.4.1. EDDS-5 original scoring (modification 1)

An initial analysis was run early in the project timeline using the original EDDS5-Dx coding, with any missing values replaced with zero. A frequency table comparing eating disorder diagnoses based upon the EDDS5-PED1 and our first validation variable (Any-SCID-ED) was created to see how well the EDDS5-PED diagnosed eating disorders as measured by SCID-enhanced interview.

2.4.2. EDDS-5 binge modified scoring (modification 2)

The diagnosis of BED resulted in the greatest number of discrepancies (i.e., misclassified cases on the EDDS-5). To meet the BINGE criterion on the EDDS-5, respondents needed to endorse items about the presence of binging (item 4), loss of control (item 5) and frequency of binges (item 6). Participants who endorsed Items 4 and 5, OR Item 6 on the EDDS-5 met criteria for a BED diagnosis on the SCID at a higher rate than if all three were required. The scoring was modified with this change to create the second diagnostic variable EDDS5-PED2. Note that the modification of the BINGE criterion scoring is involved in four diagnoses: BED, BN, low frequency BED and low frequency BN. This first scoring modification improved the sensitivity, specificity, and AUC of the overall model. However, the EDDS-5 continued to miss cases, such that the enhanced-SCID still identified five current BED diagnoses missed by the EDDS-5.

2.4.3. EDDS-5 distress modified scoring (modification 3)

The EDDS-5 scoring was reinspected for other areas of discrepancies. Endorsement of Item 12 (If you have episodes of uncontrollable overeating, does it make you very upset? YES/NO), used for BED and low frequency BED diagnoses, resulted in other misclassified cases. This item was intended to track DSM-5 impairment or distress but may have been worded in a way that did not capture this construct among veterans, resulting in misclassified cases due to observed cultural differences. More specifically, from our collective research and clinical experience we have found that veterans are not as likely to endorse being very upset by their binge episodes and that requiring this leads to underreporting of cases. Because veterans are typically have experienced a great deal of war and deployment trauma, asking about being very upset about binge eating may set a higher threshold than intended, leading to underreporting. Thus, the scoring was modified again to remove this item, resulting in the third and final diagnostic variable EDDS5-PED3.

2.4.4. EDDS-5 compensatory behavior modified scoring

The changes in the EDDS-5 scoring algorithm in this analysis were initially based on the modifications conducted by Karen Mitchell (KM) and colleagues (Mitchell et al., 2021). The original KM version used the adjusted binge frequency items (EDDS6-EDDS12) described above and removed both fasting (EDDS15) and excessive exercise (EDDS16) from compensatory behaviors. These modifications resulted in a low concordance (data not shown), which led to the need to include the modifications to the unadjusted binge frequency items (EDDS6-EDDS12) and a modified BINGE variable ((EDDS4 and EDDS5) OR EDDS6)). All these modifications together attempt to capture the fact that due to serving in the military, veterans are more likely to endorse exercising at a higher rate than civilians, especially if they are still in National Guard or Reserves and must uphold certain fitness standards. It was important to confirm that this was not what was driving the higher compensatory rate. Similarly, fasting is sometimes necessary with long tours, and intermittent fasting may be encouraged due to roles and responsibilities.

2.5. Concordance

The specificity was quite high for all diagnostic variables (averaging 0.91), but the sensitivity was quite low (averaging 0.46; see Table 3). More specifically, our ability to correctly identify people without eating disorders was quite high, but the ability to correctly identify those with eating disorders was below the acceptable range. The highest sensitivity was seen between the Current-SCID-ED+ and the EDDS5-PED3 (Sens = 0.49 (90 % CI = 0.469, 0.543); C = 0.68). The highest specificity was seen between the Any-SCID-ED+ and EDDS5-PED1 (0.964 (90 % CI = 0.949, 0.994); C = 0.72). Across the Current-SCID-ED+ concordances, sensitivity ranged from 0.41 to 0.49 and specificity ranged from 0.87 to 0.92. Similarly, for the Any-SCID-ED+ concordances, sensitivity ranged from 0.44 to 0.47 and specificity ranged from 0.89 to 0.96. See Supplemental Table S3 for unweighted cross-tabulations.

Table 3.

Weighted cross-tabulation, sensitivity, specificity, and concordance (C) with 90 % confidence intervals across each outcome compared to the original EDDS-5 (EDDS-PED1) and our modified versions (EDDS-PED2 and EDDS-PED3).

Validation variable Diagnostic variable ED− ED+ Sensitivity Specificity C
Any-SCID-ED+ EDDS5-PED1 ED− 163 84 0.44 (0.431, 0.456) 0.964 (0.949, 0.994) 0.72
ED+ 6 66
EDDS5-PED2 ED− 155 79 0.47 (0.456, 0.497) 0.919 (0.894, 0.969) 0.72
ED+ 14 70
EDDS5-PED3 ED− 151 79 0.47 (0.456, 0.497) 0.894 (0.871, 0.94) 0.70
ED+ 18 70
Current-SCID-ED+ EDDS5-PED1 ED− 193 64 0.41 (0.403, 0.424) 0.924 (0.909, 0.949) 0.66
ED+ 16 44
EDDS5-PED2 ED− 184 56 0.48 (0.462, 0.529) 0.881 (0.86, 0.918) 0.69
ED+ 25 52
EDDS5-PED3 ED− 182 55 0.49 (0.469, 0.543) 0.87 (0.851, 0.904) 0.68
ED+ 27 53
Open in a new tab

Note. SCID = Structured Clinical Interview for DSM-5; EDDS-5 = The Eating Disorder Diagnostic Scale for DSM-5; ED = eating Disorder; PED = probable eating disorder (modifications 1–3 are detailed in Results section).

3. Discussion

The EDDS-5 performed poorly at correctly identifying diverse veterans who were diagnosed with eating disorders through clinician-administered interviews. Despite multiple possible modifications tested for the EDDS-5, the sensitivity was very low, indicating that the EDDS-5 did not identify many true positives and may also over diagnose those without true eating disorders. The high specificity and relatively high concordance of the diagnostic variables may be misleading, given the low sensitivity of the EDDS-5 (0.41–0.49). Consequently, we do not recommend using the EDDS-5 for screening or diagnostic purposes among diverse samples like veterans. While self-diagnostic tools can be extremely helpful in busy healthcare systems, a new self-diagnostic tool is a critical area for future research.

Our findings highlight the importance of testing developed tools in diverse populations, given that eating disorder symptoms and disorders may manifest differently in various populations. There may be critical differences in age, gender, race, ethnicity, and trauma exposures among veterans, compared to college students. Honed measurement requires better understanding the manifestation of these disorders in diverse groups. Our findings demonstrate that the EDDS-5 may miss critical cases in veterans. Given that eating disorders are not typically screened for in existing healthcare systems, a veterans’ eating disorders may be masked or misidentified as more prevalent mental health conditions with shared symptoms. Although studies in this area are emerging, there are some investigations demonstrating that eating disorders symptoms in men being masked by depression, substance abuse and sexual abuse history (with or without PTSD; e.g., Strother, Lemberg, Stanford, & Turberville, 2012). Additionally, eating disorders can be masked as muscle bulking and over-exercising. There are several veteran studies that demonstrate high rates of comorbidity between depression, post-traumatic stress disorder, substance use and eating disorders (Maguen et al., 2012; Slane et al., 2016). Our clinical experience also has shown that eating disorders in veterans can be masked due to overweight and mental health comorbidity, which often leads to eating disorder diagnoses being delayed and diagnosed after other mental health disorders.

The EDDS-5’s poor performance may also be due to the inclusion of several new eating disorder in the DSM-5. For example, despite measurement limitations noted, AA and NES may occur at high rates among veterans (Masheb et al., 2021), and additional studies are needed to better understand how clinical manifestation as well as risk factors for these disorders may allow for clearer identification and diagnosis of eating disorders among diverse veterans. Additional reasons for poor fit may be that the EDDS-5 does not have a question about patient’s subjective binge eating, compared to clinical assessment and only a broader lifetime weight suppression question, which makes current diagnosis of AA unfeasible While the EDDS-5 was modified from the EDDS-4 to capture these disorders, the EDDS-5 has not yet been validated in any population. To our knowledge, this is the first published study to validate the EDDS-5 against gold-standard clinical interviews. Given that this study was sample-specific, further replication is needed to better understand if the EDDS-5 can accurately identify eating disorders in other populations, particularly at-risk populations.

When interpreting these findings, it is important to note that we made several modifications to the SCID Interview. The reasons for this included lack of detailed updates to this module for DSM-5, especially for BED, AA and NES – disorders with known high rates in veterans. These modifications were informed by other interviews as noted in our results. Slight modifications were also made to be more culturally attuned to veterans. We found this necessary given that in our clinical experience, men and veterans are reluctant to acknowledge some symptoms specific for these disorders. For example, acknowledging that eating is “out of control” is counter to traditional masculine characteristics, and especially military culture.

Several limitations should be noted when interpreting the findings of this study. First, our response rate was low, although not dissimilar to other studies that collected data during the COVID pandemic (de Koning et al., 2021). More specifically, de Koning et al. (2021) note that due to many studies shifting to surveys, individuals first experienced survey fatigue, followed by entirely refusing to participate in surveys at all, causing much lower response rates than have ever been seen before. Importantly, even US national and state survey response rates decreased significantly during the pandemic with an estimates 29 % decrease in response rates. This may indicate a higher rate of disordered eating behaviors than our findings suggest. Second, given that we collected data during the pandemic, there may be some variation in individuals’ eating patterns or behaviors. For example, there may have been changes in eating during the pandemic due to increased stressors or having greater access to food due to being at home. Third, veterans for the validation sample were recruited from two states where our study sites were based; for safety reasons, we also ensured that veterans were not actively suicidal and did not have situational factors that may cause their weight to fluctuate (e.g., recent childbirth); therefore, our findings may not generalize to all veterans. Fourth, while we attempted to adjust for the response rate (with regards to the interview non-response) there is currently no established method on how best to jointly adjust for non-response in both survey and interview. Future research is needed to better understand how this hierarchy of biases impacts research projects that use both qualitative and complex survey methods. Fifth, we were not able to determine if some diagnoses fit more poorly than others due to power issues, given the number of eating disorder diagnoses that are now in the DSM-5. This is an important opportunity for future studies. Future research could also focus on clinic samples with eating disorders to better understand which items on the EDDS are over or under endorsed in comparison to the SCID. We were also unable to do any sensitivity analyses by gender or any other subgroups due to our sample size. Finally, even though the EDDS-5 was not sensitive against the SCID-5, it could be that self-reported symptoms warrant healthcare attention even in the absence of a full threshold diagnosis for eating disorders.

There are several clinical implications of our study. Given that we found that the EDDS-5 had low sensitivity, we do not recommend that it be used in clinical settings with veterans as it may miss important cases or over diagnose those without true eating disorders. We also do not recommend its use in research studies with veterans for the same reasons. Although we have determined that the EDDS-5 is may not be ideal with diverse samples, there is not an easy alternative for researchers when clinical interview is not possible. Consequently, development of such a tool is an important priority for future research. For systems already using the EDDS-5, we recommend that diagnoses not be based on this measure alone, but that it be used in conjunction with additional clinical information (e.g., diagnostic interview or input from ED-experienced clinicians). More work is needed to develop screening and/or diagnostic tools with greater sensitivity and specificity that can be used across diverse populations. More research is also needed to validate the EDDS-5 in all populations. Given that this is the first known study to validate the EDDS-5 in any population, further validation studies are needed in various populations.

Supplementary Material

Paper1 Supplemental11.10

Acknowledgements

The authors would like to thank Dr. Kelly Allison for advising on measurement issues related to Night Eating Syndrome. The authors would also like to thanks Alison Marsh for her contributions to this work.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.eatbeh.2024.101846.

Footnotes

CRediT authorship contribution statement

Shira Maguen: Conceptualization, Methodology, Project administration, Writing – original draft. Adam Batten: Formal analysis, Methodology, Writing – original draft. Sarah E. Siegel: Conceptualization, Data curation, Formal analysis, Writing – review & editing. Joy Huggins: Conceptualization, Data curation, Formal analysis, Writing – review & editing. Jennifer L. Snow: Conceptualization, Writing – review & editing. Lindsay M. Fenn: Conceptualization, Writing – review & editing. Alexandra M. Dick: Writing – review & editing. Christiane Zenteno: Writing – review & editing. Anna C. West: Writing – review & editing. Robin M. Masheb: Conceptualization, Methodology, Project administration, Writing – original draft, Supervision.

Declaration of competing interest

The authors do not have any conflicts of interest to report.

1

Note: Dr. Alexandra Dick is now at McLean Hospital and Harvard Medical School in Belmont, MA and Dr. Christiana Zenteno is in private practice in San Jose, CA.

☆

Funding for this study was provided by VA HSR&D Grant IIR 17–223-2. The funders had no role in the study design, collection, analysis or interpretation of the data, writing the manuscript, or the decision to submit the paper for publication.

Data availability

The data that has been used is confidential.

References

  1. Agresti A (2012). Categorical data analysis (Vol. 792). John Wiley & Sons. [Google Scholar]
  2. Allison KC, Lundgren JD, O’Reardon JP, Martino NS, Sarwer DB, Wadden TA, … Stunkard AJ (2008). The night Eating questionnaire (NEQ): Psychometric properties of a measure of severity of the night eating syndrome. Eating Behaviors, 9(1), 62–72. https://pubmed.ncbi.nlm.nih.gov/18167324/. [DOI] [PubMed] [Google Scholar]
  3. Austin PC, & Stuart EA (2015). Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies. Statistics in Medicine, 34(28), 3661–3679. 10.1002/sim.6607 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bartlett BA, & Mitchell KS (2015). Eating disorders in military and veteran men and women: A systematic review. The International Journal of Eating Disorders, 48(8), 1057–1069. 10.1002/eat.22454 [DOI] [PubMed] [Google Scholar]
  5. Bastien CH, Vallières A, & Morin CM (2001). Validation of the insomnia severity index as an outcome measure for insomnia research. Sleep Medicine, 2(4), 297–307. 10.1016/s1389-9457(00)00065-4 [DOI] [PubMed] [Google Scholar]
  6. Begg CB, & Greenes RA (1983). Assessment of diagnostic tests when disease verification is subject to selection bias. Biometrics, 39(1), 207–215. [PubMed] [Google Scholar]
  7. Buderer NM (1996). Statistical methodology: I. Incorporating the prevalence of disease into the sample size calculation for sensitivity and specificity. Academic emergency medicine: Official journal of the society for. Academic Emergency Medicine, 3(9), 895–900. 10.1111/j.1553-2712.1996.tb03538.x [DOI] [PubMed] [Google Scholar]
  8. Cooper Z, & Fairburn C (1987). The Eating disorder examination: A semi-structured interview for the assessment of the specific psychopathology of eating disorders. International Journal of Eating Disorders, 6(1), 1–8. [DOI] [Google Scholar]
  9. Cronbach LJ (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16, 297–334. 10.1007/BF02310555 [DOI] [Google Scholar]
  10. Dorflinger LM, Ruser CB, & Masheb RM (2017). A brief screening measure for binge eating in primary care. Eating Behaviors, 26, 163–166. [DOI] [PubMed] [Google Scholar]
  11. Fairburn CG, & Beglin SJ (1994). Assessment of eating disorders: Interview or self-report questionnaire? The International Journal of Eating Disorders, 16(4), 363–370. [PubMed] [Google Scholar]
  12. Fihn SD, Francis J, Clancy C, Nielson C, Nelson K, Rumsfeld J, … Graham GL (2014). Insights from advanced analytics at the veterans health administration. Health Affairs, 33(7), 1203–1211. [DOI] [PubMed] [Google Scholar]
  13. First MB, Williams JBW, Karg RS, & Spitzer RL (2015). Structured clinical interview for DSM-5—Research version (SCID-5 for DSM-5, research version; SCID-5-RV). VA, American Psychiatric Association: Arlington. [Google Scholar]
  14. Franko DL, et al. (2012). Racial/ethnic differences in adults in randomized clinical trials of binge eating disorder. Journal of Consulting and Clinical Psychology, 80(2), 186–195. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Garner DM, Olmsted MP, Bohr Y, & Garfinkel PE (1982). The eating attitudes test: Psychometric features and clinical correlates. Psychological Medicine, 12(4), 871–878. 10.1017/s0033291700049163 [DOI] [PubMed] [Google Scholar]
  16. Gearhardt AN, Corbin WR, & Brownell KD (2009). Preliminary validation of the Yale food addiction scale. Appetite, 52(2), 430–436. 10.1016/j.appet.2008.12.003 [DOI] [PubMed] [Google Scholar]
  17. Guttman L (1945). A basis for analyzing test-retest reliability. Psychometrika, 10, 255–282. 10.1007/BF02288892 [DOI] [PubMed] [Google Scholar]
  18. Higgins DM, et al. (2013). Binge eating behavior among a national sample of overweight and obese veterans. Obesity (Silver Spring), 21(5), 900–903. [DOI] [PubMed] [Google Scholar]
  19. Hudson JI, Hiripi E, Pope HG Jr., & Kessler RC (2007). The prevalence and correlates of eating disorders in the National Comorbidity Survey Replication. Biological Psychiatry, 61(3), 348–358. 10.1016/j.biopsych.2006.03.040 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Imai K, & Ratkovic M (2014). Covariate balancing propensity score. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 76(1), 243–263. 10.1111/rssb.12027 [DOI] [Google Scholar]
  21. Kelly NR, Mitchell KS, Gow RW, Trace SE, Lydecker JA, Bair CE, & Mazzeo S (2012). An evaluation of the reliability and construct validity of eating disorder measures in white and black women. Psychological Assessment, 24(3), 608–617. 10.1037/a0026457 [DOI] [PubMed] [Google Scholar]
  22. de Koning R, Egiz A, Kotecha J, Ciuculete AC, Ooi SZY, Bankole NDA, … Kanmounye US (2021). Survey fatigue during the COVID-19 pandemic: An analysis of neurosurgery survey response rates. Frontiers in surgery, 8, Article 690680. 10.3389/fsurg.2021.690680 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Latner JD, Mond JM, Kelly MC, Haynes SN, & Hay PJ (2014). The loss of control over Eating scale: Development and psychometric evaluation. The International Journal of Eating Disorders, 47(6), 647–659. 10.1002/eat.22296 [DOI] [PubMed] [Google Scholar]
  24. Lumley T (2011). Complex surveys: A guide to analysis using R. John Wiley & Sons. [Google Scholar]
  25. Lundgren JD, Allison KC, Vinai P, & Gluck ME (2012). Assessment instruments for night eating syndrome. In Lundgren JD, Allison KC, & Stunkard AJ (Eds.), Night eating syndrome: Research, assessment, and treatment (pp. 197–217). New York: Guilford Press. [Google Scholar]
  26. Maguen S, Cohen B, Cohen G, Madden E, Bertenthal D, & Seal K (2012). Eating disorders and psychiatric comorbidity among Iraq and Afghanistan veterans. Women’s Health Issues: official publication of the Jacobs Institute of Women’s Health, 22 (4), e403–e406. [DOI] [PubMed] [Google Scholar]
  27. Masheb RM, Ramsey CM, Marsh AG, Decker SE, Maguen S, Brandt CA, & Haskell SG (2021). DSM-5 eating disorder prevalence, gender differences, and mental health associations in United States military veterans. The International Journal of Eating Disorders, 54(7), 1171–1180. 10.1002/eat.23501 [DOI] [PubMed] [Google Scholar]
  28. Mitchell KS, Masheb R, Smith BN, Kehle-Forbes S, Hardin S, & Vogt D (2021). Eating disorder measures in a sample of military veterans: A focus on gender, age, and race/ethnicity. Psychological Assessment, 33(12), 1226–1238. 10.1037/pas0001050 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Pearl RL, & Puhl RM (2014). Measuring internalized weight attitudes across body weight categories: Validation of the modified weight bias internalization scale. Body Image, 11(1), 89–92. 10.1016/j.bodyim.2013.09.005 [DOI] [PubMed] [Google Scholar]
  30. R Core Team. (2020). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from https://www.R-project.org/. [Google Scholar]
  31. Revelle W, & Condon DM (2019). Reliability from α to ω: A tutorial. Psychological Assessment, 31(12), 1395–1411. 10.1037/pas0000754 [DOI] [PubMed] [Google Scholar]
  32. Revelle WR (2017). Psych: Procedures for personality and psychological research. Software. [Google Scholar]
  33. Rudich Z, Lerman SF, Gurevich B, Weksler N, & Shahar G (2008). Patients’ self-criticism is a stronger predictor of Physician’s evaluation of prognosis than pain diagnosis or severity in chronic pain patients. The Journal of Pain, 9(3), 210–216. 10.1016/j.jpain.2007.10.013 [DOI] [PubMed] [Google Scholar]
  34. Schaefer LM, Anderson LM, Simone M, O’Connor SM, Zickgraf H, Anderson DA, … Thompson JK (2019). Gender-based differential item functioning in measures of eating pathology. The International Journal of Eating Disorders, 52(9), 1047–1051. 10.1002/eat.23126 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Selim AJ, Rogers W, Fleishman JA, Qian SX, Fincke BG, Rothendler JA, & Kazis LE (2009). Updated U.S. population standard for the veterans RAND 12-item health survey (VR-12). Quality of life research: an international journal of quality of life aspects of treatment, care and rehabilitation, 18(1), 43–52. 10.1007/s11136-008-9418-2 [DOI] [PubMed] [Google Scholar]
  36. Slane JD, Levine MD, Borrero S, Mattocks KM, Ozier AD, Silliker N, … Haskell SG (2016). Eating Behaviors: Prevalence, Psychiatric Comorbidity, and Associations With Body Mass Index Among Male and Female Iraq and Afghanistan Veterans. Military Medicine, 181(11), e1650–e1656. [DOI] [PubMed] [Google Scholar]
  37. Spitzer RL, Kroenke K, & Williams JB (1999). Validation and utility of a self-report version of PRIME-MD: the PHQ primary care study. Primary Care Evaluation of Mental Disorders. Patient Health Questionnaire. JAMA, 282(18), 1737–1744. [DOI] [PubMed] [Google Scholar]
  38. Stice E, Fisher M, & Martinez E (2004). Eating disorder diagnostic scale: Additional evidence of reliability and validity. Psychological Assessment, 16(1), 60–71. 10.1037/1040-3590.16.1.60 [DOI] [PubMed] [Google Scholar]
  39. Stice E, Telch CF, & Rizvi SL (2000). Development and validation of the eating disorder diagnostic scale: A brief self-report measure of anorexia, bulimia, and binge-eating disorder. Psychological Assessment, 12(2), 123–131. 10.1037//1040-3590.12.2.123 [DOI] [PubMed] [Google Scholar]
  40. Strother E, Lemberg R, Stanford SC, & Turberville D (2012). Eating disorders in men: Underdiagnosed, undertreated, and misunderstood. Eating Disorders, 20(5), 346–355. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Thompson-Brenner H, et al. (2013). Race/ethnicity, education, and treatment parameters as moderators and predictors of outcome in binge eating disorder. Journal of Consulting and Clinical Psychology, 81(4), 710–721. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Vogt D, Smith BN, King LA, King DW, Knight J, & Vasterling JJ (2013). Deployment risk and resilience inventory-2 (DRRI-2): An updated tool for assessing psychosocial risk and resilience factors among service members and veterans. Journal of Traumatic Stress, 26(6), 710–717. 10.1002/jts.21868 [DOI] [PubMed] [Google Scholar]
  43. Weathers FW, Litz BT, Keane TM, Palmieri PA, Marx BP, & Schnurr PP (2013). The PTSD checklist for DSM-5 (PCL-5). Scale available from the National Center for PTSD at. www.ptsd.va.gov. [Google Scholar]
  44. Williams GA, Hawkins MAW, Duncan J, Rummell CM, Perkins S, & Crowther JH (2017). Maladaptive eating behavior assessment among bariatric surgery candidates: Evaluation of the Eating disorder diagnostic scale. Surgery for obesity and related diseases: official journal of the American Society for Bariatric Surgery, 13(7), 1183–1188. 10.1016/j.soard.2017.03.002 [DOI] [PubMed] [Google Scholar]
  45. Yoshida K, Bohn J, & Yoshida MK (2020). Package ‘tableone.’ R Foundation for Statistical Computing, Vienna, Austria: (30 November 2016). [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Paper1 Supplemental11.10
NIHMS2124340-supplement-Paper1_Supplemental11_10.docx (173.2KB, docx)

Data Availability Statement

The data that has been used is confidential.

RESOURCES