SEARCH

SEARCH BY CITATION

Keywords:

  • eating disorders;
  • nosology;
  • diagnostic criteria;
  • EDNOS

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Earn CE credit for this article!
  8. References

Objective:

Revised Eating Disorder (ED) diagnostic criteria have been proposed for the Diagnostic and Statistical Manual (DSM)-5 to reduce the preponderance of eating disorder not otherwise specified (EDNOS) and increase the validity of diagnostic groups. This article compares DSM-IV and proposed DSM-5 diagnostic criteria on number of EDNOS cases and validity.

Method:

Participants (N = 397; 91% female) completed structured clinical interviews in a two-stage epidemiological study of EDs. Interviewers did not follow standard skip rules, making it possible to evaluate alternative ED diagnostic criteria.

Results:

Using DSM-IV versus DSM-5 criteria, 34 (14%) versus 48 (20%) had anorexia nervosa, 43 (18%) versus 44 (18%) had bulimia nervosa, and 163 (68%) had EDNOS versus 20 (8%) had binge eating disorder (BED), and 128 (53%) had EDNOS, respectively, reflecting a significant decrease in EDNOS. Validation analyses supported significant differences among groups with some improvement associated with delineation of BED.

Discussion:

Proposed revisions to EDs in the DSM-5 significantly reduced reliance on EDNOS without loss of information. © 2011 by Wiley Periodicals, Inc.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Earn CE credit for this article!
  8. References

Proposed diagnostic criteria for the DSM-51 include several revisions designed to reduce the preponderance of Eating Disorder Not Otherwise Specified (EDNOS) diagnoses.2 For Anorexia Nervosa (AN), amenorrhea has been eliminated as a requirement. For Bulimia Nervosa (BN), the frequency of recurrent binge-eating episodes and inappropriate compensatory behaviors has been reduced from twice to once per week over the past 3 months. The most dramatic revision involves theformal recognition of Binge ED (BED) as a specified ED.

Proposed changes were based on reviews of the extant literature published within special issues of the International Journal of Eating Disorders toward the end of 2009. Attia and Roberto3 summarized evidence indicating that amenorrhea in AN largely reflects nutritional status, suggesting that amenorrhea is a consequence of AN, similar to changes in blood pressure, body temperature, and bone mineral density, rather than a core feature of the syndrome. Furthermore, this criterion's lack of relevance for men, women taking estrogen, and premenarcheal/postmenopausal women underscored concerns about the feature's clinical utility in defining AN.3 Wilson and Sysko4 summarized evidence calling into question the validity of a twice per week minimum frequency for binge-eating episodes for either BN or BED. Only a handful of studies were available for their review because the twice weekly threshold for binge eating has been present since the DSM-III-R,5 and most studies have used DSM-based criteria in defining EDs. However, data from treatment trials suggested that a once weekly threshold could potentially reduce reliance on EDNOS without altering evidence of treatment efficacy.4 Wonderlich et al.6 reviewed evidence for the validity and clinical utility of BED. Reviewed studies supported the clinical significance of BED and distinctions between BED and AN and BN on course and outcome. In addition, psychological interventions for BED, such as cognitive behavior therapy and interpersonal therapy, demonstrated superiority to behavioral weight loss interventions for treatment of BED, suggesting that there is clinical utility in distinguishing BED from EDNOS in treatment settings.6

Although each proposed revision was based on findings from literature reviews, these findings tend to reflect the influence of changes in isolation. For example, the effect of removing amenorrhea as a criterion for AN is compared to retaining amenorrhea as a criterion rather than in combination with other proposed changes such as reducing the required binge/purge frequency for BN or considering BED as a stand-alone disorder. This reflects, in part, an inability to examine the combined effects of all possible changes before proposed revisions were put forth. However, it remains unclear whether proposed changes will have desired effects on reliance on EDNOS. In addition, it is unclear how the changes, in combination, will influence evidence of validity for diagnostic groupings.

This study compares the DSM-IV and proposed DSM-5 definitions for EDs to evaluate the success of the proposed revisions in reducing reliance on the EDNOS category. In addition, validation analyses are used to compare resulting diagnoses for the extent to which each provides clinically useful information regarding comorbid psychopathology, clinical variables, and treatment history. Data were drawn from an epidemiological study that employed a two-stage design in which structured clinical interviews were used to determine the presence of EDs among those who screened positive for an ED as well as matched controls who screened negative for an ED in the first stage. A significant advantage of this epidemiological study over other, larger and nationally representative studies (e.g., Ref. 7) was the elimination of skip rules in conducting assessments of eating pathology. This allowed comparison of different thresholds for EDs without the constraint of current definitions in detecting EDs.

Method

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Earn CE credit for this article!
  8. References

Participants

Participants (N = 397; 91% women) were recruited to complete structured clinical interviews between 2003 and 2007 as part of the second stage of a two-stage epidemiological study of EDs in college students initially recruited in 1982, 1992, and 2002.8 Importantly, this study allowed for full variation of possible ED symptoms and syndromes. Participants were between the ages of 19 and 57 years at the time of interviews; mean (SD) age was 35.4 (9.1) years. Of participants, 76.5% were Caucasian, non-Hispanic, 8.8% were Asian, 7.5% were African American, 6.8% were Hispanic, 0.3% were American Indian, and 0.3% were “Other/Mixed.”

Procedures and Measures

All participants completed written informed consent prior to participation for each stage of the two-stage epidemiological study. During the first stage, a randomly selected sample of men and women at a prestigious Northeastern University were recruited to complete self-report surveys on eating and health patterns during the springs of 1982, 1992, and 2002, with earlier cohorts recruited to complete follow-up assessments at subsequent waves of data collection. Based on reports of current ED behaviors and frequencies on surveys, cases were identified as those who met DSM-III-R criteria for a current ED for AN, BN, or EDNOS at the time of survey completion. All cases (N = 294) were invited to participate in interviews, and 196 (67%) participated. A case–control design was used to identify demographically matched controls based on age, race, and gender for each case. Controls included anyone who did not meet threshold criteria for a current ED based on self-report surveys at baseline or follow-up. However, controls could endorse disordered eating that fell below threshold for an ED diagnosis or have met criteria for an ED at some other point in their lifetime.

Interviews were conducted over the telephone by trained research assistants who were blind to participants' survey responses. Previous research has not found significant biases in conducting interviews over telephone as opposed to face to face.9, 10 To maintain blind status of interviewers, cases and controls were pursued for interviews simultaneously. When a completed case's control declined, a replacement control was identified. When a completed control's case declined, we had an unmatched control. Across the three cohorts, we had a total of 18 unmatched controls. Review of the 408 interviews revealed eight interviews (2%) in which ED status could not be determined due to interviewer error and three interviews (1%) in which participants declined to answer questions necessary to determine ED status. These were excluded from this study, resulting in a total sample of 397 participants. Because of efforts to match controls to cases on demographic variables, cases and controls did not differ in age, race, or gender (all p values > .40).

The Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I)11 modified to omit skip rules in the EDs module (e.g., participants were asked about self-induced vomiting whether or not they endorsed binge eating) was used to assess mood, anxiety, substance use, impulse control, and EDs. Interviews were audiotaped with participant consent, and 15% were randomly selected and rated by a second interviewer to assess interrater reliability. Interrater reliability, measured using κ, was 0.82 for EDs, 0.88 for mood disorders, 0.73 for anxiety disorders, 0.97 for substance use disorders, and 0.82 for impulse control disorders.

As noted above, questions that appear in Module H of the SCID were administered to all participants without following skip rules. Interviews focused on ED diagnoses during three timeframes: (1) lifetime; (2) time of survey completion; and (3) past month. This article focuses on lifetime ED diagnoses established during the structured clinical interview. Based on SCID interviews and DSM-IV criteria, 157 participants (40%) had no lifetime history of an ED (though individuals in this group could endorse isolated ED symptoms) and were used as non-ED controls in validation analyses. DSM-IV diagnostic criteria based on SCID-I interviews were used to generate DSM-IV ED diagnoses. Table 1 presents specific changes to DSM-IV criteria proposed for the DSM-5 designed to decrease use of EDNOS diagnoses that were assessed in this article. These include elimination of amenorrhea criterion for diagnosis of AN, reduction of minimum binge/inappropriate behavior frequency for a diagnosis of BN, and addition of BED as a specific type of ED with minimum frequency of binge-eating occurring once per week over a 3-month period.

Table 1. Summary of changes between DSM-IV and proposed DSM-5 diagnostic criteria for eating disorders evaluated in this study
DSM-IV Diagnostic CriteriaProposed DSM-5 Diagnostic Criteria
Anorexia nervosaAnorexia nervosa
 In postmenarcheal females, amenorrhea, i.e., the absence of at least three consecutive menstrual cycles. (A women is considered to have amenorrhea if her periods occur only following hormone, e.g., estrogen, administration.) No requirement for amenorrhea
Bulimia nervosaBulimia nervosa
 The binge eating and inappropriate compensatory behaviors both occur, on average, at least twice a week for 3 months The binge eating and inappropriate compensatory behaviors both occur, on average, at least once a week for 3 months
Eating disorder not otherwise specifiedBinge eating disorder
 The eating disorder not otherwise specified category is for disorders of eating that do not meet the criteria for any specific Eating Disorder. A. Recurrent episodes of binge eating. An episode of binge eating is characterized by both of the following:
  1. eating, in a discrete period of time (e.g., within any 2-h period),  an amount of food that is definitely larger than most people  would eat in a similar period of time under similar circumstances;
  2. a sense of lack of control over eating during the episode (e.g., a  feeling that one cannot stop eating or control what or how  much one is eating).
 B. The binge-eating episodes are associated with three (or more) of the following:
  1. eating much more rapidly than normal;
  2. eating until feeling uncomfortably full;
  3. eating large amounts of food when not feeling physically hungry;
  4. eating alone because of being embarrassed by how much one is eating;
  5. feeling disgusted with oneself, depressed, or very guilty after overeating.
 C. Marked distress regarding binge eating is present.
 D. The binge eating occurs, on average, at least once a week for 3 months.
 E. The binge eating is not associated with the recurrent use of inappropriate compensatory behaviors and does not occur exclusively during the course of bulimia nervosa or anorexia nervosa.
 Eating disorder not elsewhere classified:
 The eating disorder not elsewhere classified category is for disorders of eating that do not meet the criteria for any specific Eating Disorder.

Diagnostic categories generated by the use of DSM-IV criteria12 represent a hierarchy in which a lifetime diagnosis of AN ruled out a lifetime diagnosis of BN which ruled out a lifetime diagnosis of EDNOS (which included BED using DSM-IV criteria). Diagnostic categories generated by the use of proposed DSM-5 criteria1 also represent a hierarchy in which a lifetime diagnosis of AN ruled out a lifetime diagnosis of BN which ruled out a lifetime diagnosis of BED which ruled out a lifetime diagnosis of EDNOS. This hierarchy is consistent with diagnostic hierarchies used in other epidemiological studies13, 14 and was used to achieve parsimony in this study because the majority of ED cases (92%) were characterized by a single full-threshold diagnosis. In addition, the average number of individuals falling within various combinations of lifetime syndromes was small (n < 7). For example, despite the heuristic value of noting that an individual could migrate from meeting criteria for AN then BN and then BED over her lifetime,15 no individuals (0%) demonstrated this pattern in our sample.

Diagnostic groups were compared on educational and marital status, axis I disorders, clinical features, and treatment history information obtained from SCID-I interviews. Thus, validation analyses focused on variables that differed from the disordered eating features used to form diagnostic groups.16 Most analyses focused on nonparametric tests given the categorical nature of variables. When the omnibus test for a variable reached significance, post hoc comparisons among groups were Bonferroni adjusted for possible pairwise comparisons within each set of analyses to control family-wise error rate.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Earn CE credit for this article!
  8. References

Reduction of EDNOS from DSM-IV to Proposed DSM-5 Diagnostic Criteria

Among individuals with a lifetime history of an ED (n = 240), application of DSM-IV diagnostic criteria resulted in 34 cases (14.2%) of AN, 43 cases (17.9%) of BN, and 163 cases (67.9%) of EDNOS. In contrast, application of proposed DSM-5 diagnostic criteria resulted in 48 cases (20.0%) of AN, 44 cases (18.3%) of BN, 20 cases (8.3%) of BED, and 128 cases (53.3%) of EDNOS. Thus, proposed revisions resulted in a 20% reduction of EDNOS cases (related samples Wilcoxon signed rank test, p < .001). However, EDNOS remained the single most common eating disorder diagnosis.

Table2 depicts the migration of diagnoses between the two systems. Elimination of the amenorrhea criterion for AN resulted in a significant increase in AN diagnoses (related samples Wilcoxon signed rank Test, p < .001), most of which were drawn from the EDNOS category. In contrast, reducing the minimum frequency of episodes of binge eating and inappropriate compensatory behaviors from twice to once per week did not significantly increase the prevalence of BN (related samples Wilcoxon signed rank test, p = .655). Specifically, out of 37 EDNOS cases in which DSM-IV criteria for BN would have been met except that the frequency of binge eating and inappropriate compensatory behaviors fell below minimum thresholds, only three cases had episodes once per week for 3 months. The addition of BED as a full-threshold ED accounted for a shift of 20 cases out of the EDNOS category and was the single greatest source of reduction of EDNOS diagnoses.

Table 2. Migration of eating disorder diagnoses between DSM-IV and proposed DSM-5 diagnostic criteria
DSM-IVDSM-5
AN (n = 48)BN (n = 44)BED (n = 20)EDNOS (n = 128)
AN (n = 34)34000
BN (n = 43)24100
EDNOS (n = 163)12320128

Validation Analyses

DSM-IV Diagnostic Groups.

DSM-IV diagnostic groups demonstrated no significant differences in educational [χ2(6) = 3.95, p = .68] or marital status [χ2(3) = 3.57, p = .31]. Table3 presents results from validation analyses of DSM-IV diagnostic groups on axis I disorders, clinical features, and treatment history.

Table 3. Validation analyses of DSM-IV diagnostic categories (N = 397)
 Effect SizeNo ED (N = 157)AN (N = 34)BN (N = 43)EDNOS (N = 163)  
Validatorsϕn (%)n (%)n (%)n (%)χ2 (3)p
  • *

    Effect size for GAF analyses were computed using eta. Superscripts that differ represent significant differences between groups using a Bonferroni-corrected p value (p < .0083), except Remission (df = 2, p < .016). According to Cohen (1992), effect sizes can be characterized for Cramer's ϕ as: small = 0.10, medium = 0.30, and large = 0.50; for eta as: small = 0.10, medium = 0.25, and large = 0.40.

Lifetime axis I disorders
 Mood0.6649 (33)a23 (72)b22 (55)a,b77 (50)b22.07<.001
 Anxiety0.4018 (12)a12 (36) b8 (20)a,b38 (25)b13.45.004
 Substance use0.3825 (16)a7 (21)a,b12 (28)a,b54 (33)b13.28.004
 Impulse control0.543 (2)a7 (21)b6 (14)b17 (11)b18.09<.001
Clinical features
 Remission in past month0.35N/A25 (74)a19 (44)b101 (63)a,b7.67.02
 Suicidality0.4729 (23)a14 (50)b17 (52)b47 (33)a,b14.81.002
 GAF score [mean (SE)]0.44*79 (1)a69 (2)b,c65 (2)b71 (1)cF(3,364) = 28.91<.001
Treatment history
 Any treatment1.1372 (46)a27 (79)b38 (88)b114 (70)b39.05<.001
 Hospitalization0.313 (2)a4 (12)b2 (5)a,b3 (2)a10.65.01
 Medication0.1417 (11)6 (18)10 (23)21 (13)4.91.18

Significant group differences were found for all axis I disorders, primarily reflecting differences between individuals with no lifetime history of an ED and those with EDs. The AN and EDNOS groups reported significantly higher prevalence of mood and anxiety disorders compared to the no ED group. The EDNOS group reported significantly greater prevalence of substance use disorders compared to the no ED group, and all ED groups reported greater lifetime history of impulse control disorders compared to the no ED group. No significant differences emerged between ED groups on lifetime axis I disorders.

Remission was significantly higher in the AN group compared to the BN group, consistent with findings from a large, nationally representative epidemiological study.7 Suicidality (including thoughts of ones own death, specific plans, and suicide attempts) differed significantly across groups with higher suicidality in the AN and BN groups compared to the no ED group. Global assessment of functioning (GAF) scores were significantly worse (lower) in the ED groups compared to the no ED group. In addition, those with BN reported significantly worse psychosocial function compared to those with EDNOS. Results for GAF scores remained unchanged when controlling for remission status.

Finally, all ED groups reported greater likelihood of having received treatment for emotional or behavioral problems compared to the no ED group. The AN group, in particular, reported a greater likelihood of having received inpatient treatment compared to the no ED group and the EDNOS group.

Proposed DSM-5 Diagnostic Groups.

DSM-5 diagnostic groups demonstrated no significant differences in educational status [χ2(8) = 8.37, p = .40] but differed in marital status [χ2(4) = 10.13, p = .038]. However, no significant differences emerged between groups after adjusting for multiple comparisons. Table4 presents results from validation analyses of proposed DSM-5 diagnostic groups on axis I disorders, clinical features, and treatment history.

Table 4. Validation analyses of proposed DSM-5 diagnostic categories (N=397)
 Effect sizeNo ED (N = 157)AN (N = 48)BN (N = 44)BED (N = 20)EDNOS (N = 128)  
Validatorsϕn (%)n (%)n (%)n (%)n (%)χ2 (4)p
  • *

    Effect size for GAF analyses were computed using eta. Superscripts that differ represent significant differences between groups using a Bonferroni-corrected p value (p < .005), except remission (df = 3, p < .0083). According to Cohen (1992), effect sizes can be characterized for Cramer's ϕ as: small = 0.10, medium = 0.30, and large = 0.50; for eta as: small = 0.10, medium = 0.25, and large = 0.40.

Lifetime axis I disorders
 Mood0.6549 (33)a33 (72)b23 (56)a,b9 (47)a,b57 (48)a25.23<.001
 Anxiety0.6618 (12)a18 (38)b,c8 (20)a,b,c9 (50)c23 (19)a,b25.59<.001
 Substance use0.2925 (16)a13 (27)a,c12 (27)a,c7 (35)a,b41 (32)b,c11.65.02
 Impulse control0.423 (2)a8 (18)b6 (14)b3 (16)b13 (10)b16.40.003
Clinical features
 Remission in past month0.37N/A34 (71)20 (45)9 (45)82 (66)9.70.02
 Suicidality0.5029 (23)a20 (50)b17 (49)b8 (50)a,b33 (29)a,b18.07.001
 GAF score [mean (SE)]0.45*79 (1)a67 (1)b66 (2)b68 (2)b,c73 (1)cF(4,363) = 23.57<.001
Treatment history
 Lifetime treatment1.0072 (46)a38 (79)b39 (89)b14 (70)a,b88 (69)b39.90<.001
 Hospitalization0.283 (2)5 (11)1 (2)1 (5)2 (2)11.22.02
 Medication0.0717 (11)7 (15)9 (21)3 (16)18 (14)2.94.57

Similar to patterns with DSM-IV-based diagnostic groups, significant group effects were found for all axis I disorders reflecting higher levels in the ED groups compared to the no ED group. In addition, the AN group reported higher lifetime prevalence of mood disorders compared to the EDNOS group, and the BED group reported higher lifetime prevalence of anxiety disorders compared to the EDNOS group, despite lower power associated with smaller group sizes and more stringent Bonferroni-corrected p values.

On clinical features, a significant effect of diagnostic group remained evident for remission in the past month, and estimates were similar to those observed for groups using DSM-IV criteria. However, no significant differences emerged using Bonferroni-corrected p values. Only the AN and BN groups reported greater suicidality compared to the no ED group. All ED groups reported greater impairment compared to the no ED group, with the AN and BN groups reporting significantly worse GAF scores compared to the EDNOS group. Results were not altered by controlling for remission status.

Treatment history continued to support significant differences between the no ED group and the ED groups; however, only trend-level differences (p = .007) emerged for the pairwise comparison of hospitalization between the no ED and AN groups and between the AN and EDNOS groups. Of particular interest, despite the substantial increase (41%) in AN group size from n = 34 to n = 48 from the DSM-IV to the DSM-5 criteria, this expansion resulted in very modest changes in treatment history (i.e., 79% had ever received treatment under both definitions, and 12% vs. 11% had a history of inpatient treatment using the DSM-IV vs. DSM-5 definitions; see Tables3 and 4). Indeed, the only reason that the comparisons were significant using DSM-IV criteria and not DSM-5 criteria was the changed threshold for determining statistical significance from p < .0083 to p < .005 to reflect the addition of a fifth group for DSM-5 comparisons.

Comparison of Effect Sizes for DSM-IV Versus DSM-5 Validation Analyses.

Tables3 and 4 include effect size estimates for each validation analysis. Across comparisons, regardless of criterion set, most effect sizes fell from medium to large in magnitude. For DSM-IV definitions, large effect sizes were observed for mood and impulse control disorders, GAF scores, and history of any treatment. For proposed DSM-5 definitions, large effect sizes were observed for mood and anxiety disorders, suicidality, GAF scores, and history of any treatment. Overall, effect sizes were quite similar for both sets of definitions, suggesting very little loss of information associated with changes proposed for the DSM-5.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Earn CE credit for this article!
  8. References

The purpose of this study was to compare the DSM-IV diagnostic criteria with those recently proposed for the DSM-5 on number of EDNOS cases and validity. Results indicate that proposed revisions to diagnostic criteria for AN and the addition of BED as a separate diagnostic category will reduce the use of EDNOS among those with a lifetime ED diagnosis. However, proposed revisions to the BN criteria appear to have a very modest influence on use of the EDNOS diagnosis. Validation analyses suggested very little loss of information resulted from proposed revisions.

Given the larger number of DSM-5 groups and resulting decrease in statistical power (particularly for post hoc comparisons), one would predict greater evidence of validity for DSM-IV diagnoses compared to DSM-5 diagnoses even if no true differences existed. However, significant differences emerged in lifetime history of mood and anxiety disorders and GAF scores among ED groups for DSM-5 criteria that were not observed for DSM-IV criteria. Thus, findings suggest slightly improved discrimination among ED groups with the expansion of AN to include those who menstruate and separation of BED from other EDNOS. In addition, comparison of lifetime axis I disorders, clinical features, and treatment history between DSM-IV AN and DSM-5 AN (see Tables 3 and 4) indicates no evidence that the expansion of AN diluted the severity of illness in this group. If anything, remission in the past month decreased, and GAF scores worsened slightly (with slightly decreased standard errors) with the category's expansion.

The limited impact of proposed revisions to the BN diagnostic criteria may reflect the narrow band of individuals who binge and compensate at least once a week but not twice a week over a 3-month period. This finding is consistent with results from a clinic-based sample17 noted by Wilson and Sysko in their review.4 Our review of EDNOS cases that failed to meet the frequency/duration criterion for BN suggest that less conservative adjustments would be required for both the frequency and duration criteria to significantly reduce cases of EDNOS resembling subthreshold BN. However, such changes could result in significant loss of information. In addition to the narrow band of individuals who may be accommodated by proposed changes to BN criteria, some individuals who received a lifetime diagnosis of BN using the DSM-IV criteria were reassigned to lifetime diagnoses of AN due to expansion of the latter group. The overall effect was that the BN group was approximately the same size using DSM-IV or DSM-5 criteria. The similarity in axis I disorders, clinical features, and treatment history for BN defined by DSM-IV versus DSM-5 criteria suggests very little loss of information with proposed revisions.

Finally, the proposed addition of BED resulted in the single largest decrease in EDNOS cases. Validation analyses supported its distinction from other EDNOS on the basis of higher lifetime history of anxiety disorders in BED and detection of a significant difference in lifetime mood disorders between AN and EDNOS after BED was removed from the latter category. The BED group endorsed levels of suicidality that were comparable to those reported by the AN and BN groups (50%, 50%, and 49%, respectively). However, this did not emerge as a significant difference from EDNOS, potentially reflecting limited statistical power for this comparison.

The current results have several clinical implications. First, if the DSM-5 adopts proposed changes to diagnostic criteria for AN and BN, this will result in very little loss of information, consistent with the bases on which changes were proposed.3, 4, 6 For example, the amenorrhea criterion does not appear to add to the concurrent validity of AN diagnoses and thus does not appear to be a valid basis upon which to distinguish between cases of AN and EDNOS. However, clinicians may still be faced with patients who do not meet new criteria for BN because their episodes have persisted for only 2 months or occur less frequently than once per week. This seems particularly likely as treatment history did not distinguish between those with either DSM-IV or proposed DSM-5 diagnoses of BN and EDNOS. Finally, the ability to distinguish individuals with BED from the broader category of EDNOS may help clinicians identify likely differences in comorbid axis I disorders which appear to be lower in EDNOS, once BED is removed. Larger sample sizes are needed to determine whether apparent differences in remission and suicidality also distinguish BED from EDNOS.

This study has some notable strengths and limitations. First, data came from an epidemiological study that used structured clinical interviews but did not adhere to skip rules when assessing eating pathology. Thus, symptom-level ED data were available for all participants. Second, data did not come from treatment-seeking samples which protects against inflated estimates of comorbid axis I disorders18 that could mask potential differences among EDs in comorbidity. Third, stringent alpha values were used to decrease the risk of type I error, and effect sizes were included to allow evaluation of differences that emerged in validation analyses independent of changes related to group size and number. Fourth, this study evaluated a range of variables to establish concurrent validity of diagnoses, including comorbid disorders, GAF, suicidality, remission, and treatment history—all of which represent important clinical indicators.

One limitation is that the sample sizes were not large enough to detect small effect sizes for group comparisons, which contributed to restricted statistical power during pairwise comparisons of groups with post hoc-corrected p values. Relatedly, we did not examine AN and BN subtypes in migration or validation analyses given that this approach would have further reduced group sizes. Given the proposal to eliminate the nonpurging subtype of BN, this is an important topic for future investigations. In addition to not evaluating proposed changes in subtypes, we were unable to evaluate the influence of proposed clarifications in the wording of certain AN criteria. For example, the proposed change for criterion A of AN does not present an alternative weight threshold that we could evaluate. Moreover, this criterion may be established using the example threshold provided in the DSM-IV (as we did in our analyses), using a less stringent threshold, or using a more stringent threshold. Similarly, we were unable to reconstruct from interviews “persistent behavior to avoid weight gain” (proposed revision to criterion B of AN as an alternative to explicit endorsement of intense fear of gaining weight or becoming fat). Thus, our estimate of the increase in AN diagnoses with proposed DSM-5 criteria is likely to be conservative. Finally, our validation analyses focused exclusively on variables that represent concurrent validity—that is, assessment of other lifetime variables assessed concurrently with assessment of lifetime ED diagnoses. Future studies should examine whether proposed changes influence evidence of etiologic and predictive validity.19

Overall, findings support proposed revisions toEDs in the DSM-5 as successfully reducing reliance on EDNOS without significant loss of information. The most controversial change, introduction of BED, did not result in an abundance of this diagnosis in this community-based sample. Moreover, proposed revisions reflect a fairly conservative approach to changing ED definitions similar to that employed in revising previous editions of the DSM. The key advantage of this conservative approach, as evidenced by the current findings, is very little loss of information. The key disadvantage is that it appears likely that EDNOS may continue to be the most common diagnosis, particularly in community-based samples. However, proposals to include specific, named alternative forms of EDNOS in the DSM-5,1 such as purging disorder and night eating syndrome, may pave the way for continued progress in future editions of the DSM.

Earn CE credit for this article!

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Earn CE credit for this article!
  8. References

Visit: http://www.ce-credit.com for additional information. There may be a delay in the posting of the article, so continue to check back and look for the section on Eating Disorders. Additional information about the program is available at www.aedweb.org

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Method
  5. Results
  6. Discussion
  7. Earn CE credit for this article!
  8. References