Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Int J Eat Disord. Author manuscript; available in PMC 2010 November 1.
Published in final edited form as:
PMCID: PMC2844095

Validity and Utility of Subtyping Anorexia Nervosa

Christine Peat, B.A.,1 James E. Mitchell, M.D.,2,* Hans Hoek, M.D., Ph.D.,3 and Stephen Wonderlich, Ph.D.2



The purpose of this paper is to review the available literature that addresses the predictive validity and utility of subtyping anorexia nervosa patients into binge/purge and restrictor subtypes.


Literature was reviewed including studies that compared individuals with subtype diagnoses on clinical and outcome variables as well as more recent research examining the frequency of diagnostic crossover.


Several studies found that in general the binge/purge subtype patients have more psychopathology, tend to be older and tend to have a worse outcome. More recent studies which have examined diagnostic crossover suggest that the rate of crossover from the restrictor subtype to the binge/purge subtype is substantial. Crossover from the binge/purge to the restrictor subtype appears to occur less commonly. There is also literature documenting crossover from anorexia nervosa to bulimia nervosa and a small literature looking at crossover from bulimia nervosa to anorexia nervosa.


The results of this review suggest that although there is generally progression from restrictor anorexia nervosa to binge/purge anorexia nervosa to bulimia nervosa in a sizeable number of patients, other crossover patterns can be seen as well, and the amount of crossover is quite large. This suggests a lack of predictive validity for subtypes. Validity and Utility of Subtyping Anorexia Nervosa


There has been a growing interest in questions concerning the diagnoses of psychiatric disorders since the recent initiation of planning for the development of the DSM-V. As part of this process certain questions have been identified as important when considering possible modifications to the criteria for eating disorders. One such question concerns the subtyping of anorexia nervosa (AN) into the restrictor versus binge/purge subtypes currently employed in DSM-IV. Several questions can be posed regarding the utility of this subtyping approach, including: 1) Do clinicians use the subtyping scheme appropriately? 2) Do the data support the predictive validity of the subtyping? 3) Are genetic, taxometric and pharmacotherapy response data consistent with subtyping? and 4) Does the subtyping scheme provide clinically useful information for treatment planning?

All four of these questions are relevant in deciding whether or not the AN subtyping scheme should be maintained as is, eliminated or amended in the next version of the DSM. The purpose of this paper is to consider each of these questions by examining the relevant literature, as well as by considering the clinical implications of these questions where empirical data do not exist.


In order to examine the validity and clinical utility of AN subtyping, we conducted a comprehensive literature review. We searched major computer databases (e.g., MedLine, PsychInfo) and also reviewed reference lists from published literature. Search terms included anorexia nervosa, anorexia nervosa binge eat, anorexia nervosa binge purge, anorexia nervosa purge.


I. Do clinicians use the subtyping scheme appropriately?

We were unable to find any literature that directly addressed this question. Consideration might be given to a field trial to address this question.

II. Do the data support the predictive validity of the subtypes?

Early reports, as well as some more recent work, suggest that the binge/purge or bulimic subgroup of patients with AN differed in significant ways from patients with the restrictor form of AN (Table 1).

Table 1
Clinical Correlates of AN Subtype Diagnoses

Halmi et al.1 reported that purging behavior in patients with AN predicted a more negative outcome. Two classic papers appeared in 1980. Casper et al., reported data on 105 hospitalized patients from a multi-center treatment study of AN2. Forty-seven percent had periodic binge eating and purging and 53% were consistently restrictors. ANBP patients were found to be more extroverted, reported a stronger appetite, tended to be older, and also more commonly reported kleptomania, as well as greater levels of anxiety, depression, and guilt. The same year Garfinkel et al. 3 reported on 141 AN patients, 68 of whom had bulimic symptoms and 73 of whom were restrictors. They reported that the ANBP subgroup had a history of weighing more and were more commonly premorbidly obese. The bulimic subgroup more commonly reported a variety of impulsive behaviors including use of alcohol and street drugs, stealing, suicide attempts and self-injurious behavior. They found a positive family history of obesity among the ANBP subgroup. Both papers argued for a subtyping scheme. Laessle et al. in 19894 reported that in a sample of 41 AN patients the restrictor subgroup had lower rates of comorbidity than the ANBP subgroup. Garner et al. in 19935 reported results of a study examining 390 patients with AN cross-sectionally and concluded that those who binge ate or purged had more psychopathology than the restrictor group. Deter and Herzog in 19946 reported on a sample of 84 AN patients from Heidelberg who were followed up for 12 years. They found that purging predicted a negative outcome in this prospective study.

Favaro and Santonastaso in 19957 reported on 164 AN patients, finding that those who used multiple purging methods had greater levels of impulsivity and were more likely to report self-injurious behavior. Pryor et al. in 19968 reported on a sample of 171 AN patients finding that the binge/purge subtype reported more stealing behavior and more suicide attempts. However, in contrast to many of the other reports, Herzog et al. in 19969 reported on a sample of 75 AN patients from the Boston area followed prospectively and found that the ANBP group was more likely to recover than the ANR group. However, in the Heidelberg sample, Herzog et al. in 199710 reported that, in 69 AN subjects at 12 year follow-up, restriction predicted earlier recovery. More recently Godart et al. in 200611 reported on a sample of 166 AN subjects and found equal rates of anxiety and depression in the AN restrictor and binge/purge subgroups.

In summary, with a few exceptions, binge/purging cross-sectionally appears to be associated with higher rates of impulsivity, more reports of stealing behavior, suicide attempts and self-injurious behavior and, in all but one of the prospective studies, appears to be associated with a more negative outcome.

Literature on subtyping has increasingly focused on the issue of diagnostic crossover (Table 2). However, in considering the literature it is important to keep in mind the descriptions as to what constitutes these subtypes have been inconsistent. Some of this literature has focused on crossover between the binge/purge and restrictor subtypes. Eckert et al. in 199512 in a report concerning a ten-year follow-up of 76 patients reported that the crossover rate from ANR to ANBP was 64%. Strober et al. in 199713 reporting on a sample of 95 patients followed up for 10 to 15 years found a crossover rate from ANR to ANBP of 30%. In the Boston sample, Eddy et al. in 200214 reported a crossover rate from ANR to ANBP of 62% and in their 2008 paper reported a crossover rate from ANR to ANBP of 55%16. Fichter and Quadflieg in 200615 reported on a twelve-year follow-up on 311 AN patients and found a crossover rate from ANR to ANBP of 35%. Anderluh et al. found a crossover rate from ANR to ANBP of 38%17. However, crossover can also occur from ANBP to ANR. Fichter and Quadflieg found a crossover rate from ANBP to ANR of 18% in a sample of 31115, and Eddy et al. in 200814 reported a crossover rate from ANBP to ANR of 44%. Anderluh et al. reported a crossover rates from ANBP to ANR of 17%17.

Table 2
Studies of Diagnostic Crossover from AN R to AN BP and fromAN BP to AN R

Crossover can also occur from AN to BN (Table 3). Milos, et al., in 2005 in a sample of 192 treatment seeking patients found that at 30-month follow-up 20% had crossed over from AN to BN18. Fichter, et al., in 200615 at a twelve-year follow-up of 97 AN patients found that 10% had crossed over to BN and Eddy, et al., in 2008 in a sample of 216 found that 10% of the ANR and 54% of the ANBP had crossed over to BN16. Anderluh, et al. found a crossover rate from ANBP to BN of 29%17.

Table 3
Studies of Diagnostic Crossover from AN to BN

Lastly crossover from BN to AN has also been reported (Table 4). Fairburn et al.19 reported a 5-year follow-up on a community sample of 92 subjects. They reported that only 2% of patients with BN received a diagnosis of AN over 5 years. Milos et al. 18 in a treated sample of 192 at 30-month follow-up found a BN to AN crossover rate of 9%. However, Tozzi, et al. 20, in a retrospective study of 216 patients recruited as part of a genetic sample found a crossover rate over seven years of 27% from BN to AN. Anderluh et al. found a crossover rate from BN to AN of 17%17.

Table 4
Studies of Diagnostic Crossover from BN to AN

In sum, this literature suggests that, in well-characterized samples followed for intermediate to long periods of time, there is significant crossover between diagnostic subgroups of AN, from AN to BN and some crossover from BN to AN as well.

III. Are genetic, taxometric and pharmacotherapy response data informative on this question?

One consideration is the examination for susceptibility genes for AN. A genome-wide linkage analysis of 192 families with at least one affected relative pair resulted in only modest evidence for linkage, with the highest nonparametric score of 1.80 at marker D4S2367 on chromosome 4. Subsequently linkage analysis was done in a subset of families in which at least two affected relatives had a diagnosis of ANR. The NPL nonparametric linkage score observed was 3.03 at marker D1S3721 on chromosome 121. Subsequently genotyping of 9 single nucleotide polymorphisms suggested that genes for the serotonin 1D receptor and the opioid delta receptor or linked genes might be involved in the risk for restrictor AN22. This suggests the possibility that genetic differences exist between AN subtypes.

Taxometric studies of eating disorder participants have also been published. In relation to the restricting subtype of AN, Williamson, et al., in 2002 reported that ANBP appeared to occur on a continuum with normality23. Both Gleaves et al. 24 and Williamson et al. 23 found that the ANR subtype was qualitatively different from the ANBP subtype, finding that the ANBP subtype showed evidence of continuity with BN. This paralleled the findings from latent class analysis that placed BN and ANBP in the same class25.

Other research suggests that those with ANBP do not respond to certain treatments to which BN patients respond such as fluoxetine26. Two studies, however, have found differential responses between those with ANR and ANBP to other pharmacotherapies, including cyproheptadine and an atypical antipsychotic27, 28.

IV. Does the subtyping scheme provide useful clinical information for the practitioner that can be used in treatment planning?

While it seems reasonable that clinicians may make use of subtype information in assessing patients and in treatment planning, we were unable to locate any empirical data directly addressing this issue. Again, this might be a focus for a field trial.


There appears to be lack of predictive validity for the AN subtypes. There is often progression from ANR to ANBP and from ANBP to BN, but other crossover patterns occur as well. In general data are lacking concerning whether clinicians use the subtyping scheme appropriately, and whether the subtypes provide useful information for practitioners, although the assumption might be that they do. Data on genetic, taxometric and pharmacotherapy response again are inconsistent.

It is not clear that the definitions offered in the DSM-IV allow clinicians to reliably distinguish between the restrictor versus binge-purge subtype. One consideration is that the criterion for restricting subtype stipulates that those in this subgroup will not have “regularly engaged” in binge eating or purging behavior. How “regularly engaged” is best defined is unclear, and no cut points are offered in DSM-IV TR. Therefore, clinicians are free to interpret this criterion as they wish, and there are no data suggested how this is usually done in practice. By extension, it is also unclear what period of time must elapse before a subtype can be considered to have changed, given that it is clear that patients can change subtype diagnoses over time, as will be discussed further. Deciding on the presence or absence of binge eating and purging is probably particularly problematic when patients engage in such behaviors at a low frequency.

Another point of consideration is a lack of clarity as to what constitutes binge eating in AN. Clinical lore suggests that many patients with AN eat relatively modest amounts of food when they indicate they had an eating binge. An example would be a patient who states she has been binge eating, but when asked what she actually consumed, indicates that she ingested a cookie. This amount, and probably the macronutrient content of this food, indicated to her that the amount she had taken in was “excessive” or “forbidden”. However, there are data indicating that eating binges in those with AN can be quite large29. Thus there may be considerable variability in size of binge eating episodes in patients with AN, which is in line with what we know from feeding laboratory studies in those with bulimia nervosa (BN) and binge eating disorder (BED), where size of binge eating episodes varies dramatically30, 31.

In relation to predictive validity, the data are to some extent clearer. Most research in this area suggests that an individual’s diagnostic subtype designation may change over time and that the most common patterns of crossover are from restrictor type AN to binge-purge type AN, and from binge-purge AN to BN. However, sizeable subgroups of patients move between subtypes and diagnoses in other ways and move in to and out of subsyndromal states as well. Therefore, subtype diagnoses determined cross-sectionally do not appear to offer strong predictive validity about course or outcome.

This literature review suggests several possible options:

  1. One option is to continue the current system. This would provide continuity with the DSM-IV-TR diagnostic nomenclature. The clear advantage of this would be to minimize change, given prior research findings and that many clinicians are familiar with the current subtyping scheme. Also, subtyping logically seems to have potential clinical utility in that it should be of some use in assessment and treatment planning. Retaining the subtyping scheme may also facilitate differential treatment research. The main argument against this choice is the lack of predictive validity of the diagnostic subtypes, and the problem of diagnostic crossover. Another consideration is that if the subtyping scheme is to be retained, more precise guidelines might be included regarding the necessary frequency and duration of binge eating and purging required for that subtype diagnosis. However, we could find no data that empirically addressed this issue, and the definitions used in the literature in terms of frequency of behaviors have varied considerably.
  2. A second possibility would be to alter the subtyping scheme for AN to emphasize that it indicates whether they are currently binge/purge or restrictor. This scheme would retain some of the advantages of the current system but the use of the term “currently” would underscore that diagnostic crossover may well have occurred in the past or may occur later. This would be novel and might be confusing to clinicians.
  3. A third approach would be to simply eliminate the subtyping scheme for AN. This would be in keeping with the lack of predictive validity but would make more difficult the communication of subtyping information which may be important in assessment and treatment planning.


Supported by NIH R13 MH177443


1. Halmi K, Brodland G, Loney J. Prognosis in anorexia nervosa. Annals of Int Med. 1973;78:907–909. [PubMed]
2. Garfinkel PE, Moldofsky H, Garner DM. The heterogeneity of anorexia nervosa. Bulimia as a distinct subgroup. Arch Gen Psychiatry. 1980;37:1036–1040. [PubMed]
3. Casper RC, Eckert ED, Halmi KA, Goldberg SC, Davis JM. Bulimia. Its incidence and clinical importance in patients with anorexia nervosa. Arch Gen Psychiatry. 1980;37:1030–1035. [PubMed]
4. Laessle RG, Wittchen HU, Fichter MM, Pirke KM. The significance of subgroups of bulimia and anorexia nervosa: Lifetime frequency of psychiatric disorders. Int J Eat Disord. 1989;8:569–574.
5. Garner DM, Garner MV, Rosen LW. Anorexia nervosa “restricters” who purge: Implications for subtyping anorexia nervosa. Int J Eat Disord. 1993;13:171–185. [PubMed]
6. Deter H, Herzog W. Anorexia nervosa in a long-term perspective: Results of the Heidelberg-Mannheim study. Psychosom Med. 1994;56:20–27. [PubMed]
7. Favaro A, Santonastaso P. Purging behaviors, suicide attempts, and psychiatric symptoms in 398 eating disordered subjects. Int J Eat Disord. 1996;20:99–103. [PubMed]
8. Pryor T, Wiederman MW, McGilley B. Clinical correlates of anorexia nervosa subtypes. Int J Eat Disord. 1996;19:371–379. [PubMed]
9. Herzog D, Field AE, Keller MB, West JC, Robbins WM, Staley J, Colditz GA. Subtyping eating disorders: Is it justified? J Am Acad Child Adolesc Psychiatry. 1996;35:7. [PubMed]
10. Herzog W, Schellberg D, Deter H. First recovery in anorexia nervosa patients in the long-term course: A discrete-time survival analysis. J Consult Clin Psychol. 1997;65:169–177. [PubMed]
11. Godart N, Berthoz S, Rein A, Perdereau F, Lang F, Venisse J, Halfon O, Bizouard P, Loas G, Corcos M, Jeammet P, Flament M, Curt F. Does the frequency of anxiety and depressive disorders differ between diagnostic subtypes of anorexia nervosa and bulimia? Int J Eat Disord. 2006;39:772–778. [PubMed]
12. Eckert ED, Halmi KA, Marchi P, Grove W, Crosby R. Ten-year follow-up of anorexia nervosa: clinical course and outcome. Psychol Med. 1995;25:143–156. [PubMed]
13. Strober M, Freeman R, Morrell W. The long-term course of severe anorexia nervosa in adolescents: Survival analysis of recovery, relapse, and outcome predictors over 10–15 years in a prospective study. Int J Eat Disord. 1997;22:339–360. [PubMed]
14. Eddy KT, Keel PK, Dorer DJ, Delinsky SS, Franko DK, Herzog DB. Longitudinal comparison of anorexia nervosa subtypes. Int J Eat Disord. 2002;31:191–201. [PubMed]
15. Fichter MM, Quadflieg N, Hedlund S. Twelve-year course and outcome predictors of anorexia nervosa. Int J Eat Disord. 2006;39:87–100. [PubMed]
16. Eddy KT, Dorer DH, Franko DL, Tahilani K, Thompson-Brenner H, Herzog DB. Diagnostic crossover in anorexia nervosa and bulimia nervosa: Implications for DSM-V. Am J Psychiatry. 2008;165:2. [PMC free article] [PubMed]
17. Anderluh M, Tchanturia K, Rabe-Hesketh S, Collier D, Treasure J. Lifetime course of eating disorders: design and validity testing of a new strategy to ddefine the eating disorders phenotype. Psychol Med. 2008 [PubMed]
18. Milos G, Spindler A, Schnyder U, Fairburn CG. Instability of eating disorder diagnoses: prospective study. Br J of Psychiatry. 2005;187:573–578. [PMC free article] [PubMed]
19. Fairburn CG, Cooper Z, Doll HA, Norman P, O’Connor M. The natural course of bulimia nervosa and binge eating disorder in young women. Arch Gen Psychiatry. 2000;57:659–665. [PubMed]
20. Tozzi F, Thornton LM, Klump KL, Fichter MM, Halmi KA, Kaplan AS, Strober M, Woodside DB, Crow S, Mitchell J, Rotondo A, Mauri M, Cassano G, Keel P, Plotnicov KH, Pollice C, Lilenfeld LR, Berrettini WH, Bulik CM, Kaye WH. Symptom fluctuation in eating disorders: Correlates of diagnostic crossover. Am J Psychiatry. 2005;162:4. [PubMed]
21. Grice DE, Halmi KA, Fichter MM, Strober M, Woodside DB, Treasure JT, Kaplan AS, Magistretti PJ, Goldman D, Bulik CM, Kaye WH, Berrettini WH. Evidence for sa susceptibility gene for anorexia nervosa on chromosome 1. Am J Hum Genet. 2002;70:787–792. [PubMed]
22. Bergen AW, van den Bree MBM, Yeager M, Welch R, Ganjei JK, Haque K, Bacanu S, Berrettini WH, Grice DE, Goldman D, Bulik CM, Klump K, Fichter M, Halmi K, Kaplan A, Strober M, Treasure J, Woodside B, Kaye WH. Candidate genes for anorexia nervosa in the 1p33–36 linkage region: serotonin 1D and delta opioid receptor loci exhibit significant association to anorexia nervosa. Mol Psychiatry. 2003;8:397–406. [PubMed]
23. Williamson DA, Womble LG, Smeets MA, Netmeyer RG, Thaw JM, Kutlesic V, Gleaves DH. Latent structure of eating disorder symptoms: A factor analytic and taxometric investigation. Am J Psychiatry. 2002;159:412–418. [PubMed]
24. Gleaves D, Lowe MR, Green BA, Cororve MB, Williams TL. Do anorexia and bulimia nervosa occur on a continuum? A taxometric analysis. Beh Therapy. 2000;31:195–219.
25. Wade TD, Crosby RD, Martin NG. Use of latent profile analysis to identify eating disorder phenotypes in an adult Australian twin cohort. Arch Gen Psychiatry. 2006;63:1377–1384. [PubMed]
26. Walsh BT, Kaplan AS, Attia E, Olmsted M, Parides M, Carter JC, et al. Fluoxetine after weight restoration in anorexia nervosa: a randomized controlled trial. JAMA. 2006;295:2605–2612. [PubMed]
27. Halmi KA, Eckert E, LaDu TJ, Cohen J. Anorexia nervosa: Treatment efficacy of cyproheptadine and amitriptyline. Arch Gen Psychiatry. 1986;43:177–181. [PubMed]
28. Brambilla F, Barcia CS, Fassino S, Daga GA, Favaro A, Santonestoso P, et al. Olanzapine therapy in anorexia nervosa: psychobiological effects. Int Clin Psychopharm. 2007;22:197–204. [PubMed]
29. Burd C, Mitchell JE, Crosby RD, Engel SG, Wonderlich SA, Lystad C, le Grange D, Peterson CB, Crow S. An assessment of food intake in anorexia nervosa subjects in the natural environment. Paper presented at the Eating Disorders Research Society Meeting; Montreal, Canada. 9/08.
30. Samuels F, Zimmerli EJ, Devlin MJ, Kissileff HR, Walsh BT. The development of hunger and fullness during a laboratory meal in patients with binge eating disorder. Int J Eat Disorder. 2007 (ahead of print) [PMC free article] [PubMed]
31. Kissileff HR, Zimmerli EJ, Torres MI, Devlin MJ, Walsh BT. Effect of eating rate on binge size in Bulimia Nervosa. Psyciol Behav. 2008;93:481–485. [PMC free article] [PubMed]