specific and nonspecific comorbidity in anorexia nervosa

10
Specific and Nonspecific Comorbidity in Anorexia Nervosa Jennifer Jordan, PhD 1 * Peter R. Joyce, MD, PhD, DSc, FRSNZ, FRANZCP 1 Frances A. Carter, PhD 1 Jacqueline Horn, PhD 2 Virginia V.W. McIntosh, PhD 1 Suzanne E. Luty, BM, BS, PhD, FRANZCP 1 Janice M. McKenzie, MB, ChB, FRANZCP 1 Christopher M.A. Frampton, PhD 1 Roger T. Mulder, MB ChB, PhD, FRANZCP 1 Cynthia M. Bulik, PhD 3,4 ABSTRACT Objective: This article reports lifetime Axis I and II comorbidity in women with anorexia nervosa (AN), and ascertains specific and nonspecific comorbidity in AN compared to clinical samples of women with bulimia nervosa (BN) or major depression (DEP). Method: Outpatient AN (n 5 56), BN (n 5 132), and DEP (n 5 100) samples were assessed using Structured Clinical Inter- views I and II for DSM-III-R. Baseline data were compared using univariate statistics and logistic regression. Results: In the AN sample as a whole, specific elevations were found for preva- lences of obsessive compulsive disorder. The AN-binge eating purging subtype (AN-BP) and the BN sample had elevated prevalences of Cluster B personality dis- orders. Cluster C prevalences were ele- vated across samples. Conclusion: Evidence of AN-specific, eating disorder-specific, and nonspecific comorbidity illustrates the heterogeneity in AN. Further research is need to exam- ine the relative impact of specific and nonspecific comorbidity in AN subtypes and AN as a whole. V V C 2007 by Wiley Peri- odicals, Inc. Keywords: cormorbidity; anorexia ner- vosa; bulimia nervosa; major de-pressive disorder; anxiety; psychoactive substance use; personality disorder (Int J Eat Disord 2008; 41:47–56) Introduction Heterogeneity appears to be the rule rather than the exception in those with eating disorders, including anorexia nervosa (AN). One of the impor- tant ways that individuals with AN often vary is in the presence or absence of particular comorbid psychiatric diagnoses. Considerable interest exists in using a comorbidity approach to identify mean- ingful subgroups among those with eating disor- ders (e.g. Refs. 1–3). Psychiatric comorbidity is very common in epi- demiologic studies 4,5 and reported prevalences of comorbid diagnoses tend to be higher in clinical samples. 6 Greater comorbidity in clinical samples has also been found with eating disorders (e.g., Ref. 7) although this observation is not universal. 8 Although numerous studies have described preva- lences of comorbid disorders in AN, there has been less attention paid to whether the observed comor- bidity represents specific elevations in AN com- pared to other clinical comparison groups, or whether they may be nonspecific—that is, they may be similarly elevated in those presenting for treat- ment of other psychiatric disorders. The literature on the impact of comorbidity on outcome in AN is mixed. Some studies report no impact of comorbidity 9,10 ; however, other studies report poorer outcome when comorbid disorders are present. 11–17 The impact of comorbidity may be on the individual’s general psychosocial function- ing, rather than on the eating disorder directly. 18,19 Concurrent comorbidity has been reported to be more pronounced in the acute phase of AN, with depression and obsessionality either secondary to or exacerbated by the starvation state. 20,21 Improvement in comorbid symptoms often ac- companies weight restoration, however persisting Accepted 15 July 2007 1 Department of Psychological Medicine, University of Otago, Christchurch School of Medicine and Health Sciences, Christchurch, New Zealand 2 Department of Respiratory Medicine, Christchurch Hospital, Canterbury District Health Board, Christchurch, New Zealand 3 Department of Psychiatry, University of North Carolina at Chapel Hill, North Carolina 4 Department of Nutrition, University of North Carolina at Chapel Hill, North Carolina *Correspondence to: Jennifer Jordan, Department of Psychological Medicine, University of Otago, Christchurch School of Medicine and Health Sciences, PO Box 4345, Christchurch 8140, New Zealand. E-mail: [email protected] Supported from Health Research Council of New Zealand. Published online 14 September 2007 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/eat.20463 V V C 2007 Wiley Periodicals, Inc. International Journal of Eating Disorders 41:1 47–56 2008—DOI 10.1002/eat 47 REGULAR ARTICLE

Upload: jennifer-jordan

Post on 11-Jun-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

Specific and Nonspecific Comorbidity inAnorexia Nervosa

Jennifer Jordan, PhD1*Peter R. Joyce, MD, PhD, DSc,FRSNZ, FRANZCP1

Frances A. Carter, PhD1

Jacqueline Horn, PhD2

Virginia V.W. McIntosh, PhD1

Suzanne E. Luty, BM, BS, PhD,FRANZCP1

Janice M. McKenzie, MB, ChB,FRANZCP1

Christopher M.A. Frampton,PhD1

Roger T. Mulder, MB ChB, PhD,FRANZCP1

Cynthia M. Bulik, PhD3,4

ABSTRACT

Objective: This article reports lifetime

Axis I and II comorbidity in women with

anorexia nervosa (AN), and ascertains

specific and nonspecific comorbidity in

AN compared to clinical samples of

women with bulimia nervosa (BN) or

major depression (DEP).

Method: Outpatient AN (n 5 56), BN (n

5 132), and DEP (n 5 100) samples were

assessed using Structured Clinical Inter-

views I and II for DSM-III-R. Baseline data

were compared using univariate statistics

and logistic regression.

Results: In the AN sample as a whole,

specific elevations were found for preva-

lences of obsessive compulsive disorder.

The AN-binge eating purging subtype

(AN-BP) and the BN sample had elevated

prevalences of Cluster B personality dis-

orders. Cluster C prevalences were ele-

vated across samples.

Conclusion: Evidence of AN-specific,

eating disorder-specific, and nonspecific

comorbidity illustrates the heterogeneity

in AN. Further research is need to exam-

ine the relative impact of specific and

nonspecific comorbidity in AN subtypes

and AN as a whole. VVC 2007 by Wiley Peri-

odicals, Inc.

Keywords: cormorbidity; anorexia ner-

vosa; bulimia nervosa; major de-pressive

disorder; anxiety; psychoactive substance

use; personality disorder

(Int J Eat Disord 2008; 41:47–56)

Introduction

Heterogeneity appears to be the rule rather thanthe exception in those with eating disorders,including anorexia nervosa (AN). One of the impor-tant ways that individuals with AN often vary is inthe presence or absence of particular comorbidpsychiatric diagnoses. Considerable interest existsin using a comorbidity approach to identify mean-ingful subgroups among those with eating disor-ders (e.g. Refs. 1–3).

Psychiatric comorbidity is very common in epi-demiologic studies4,5 and reported prevalences ofcomorbid diagnoses tend to be higher in clinicalsamples.6 Greater comorbidity in clinical sampleshas also been found with eating disorders (e.g.,Ref. 7) although this observation is not universal.8

Although numerous studies have described preva-lences of comorbid disorders in AN, there has beenless attention paid to whether the observed comor-bidity represents specific elevations in AN com-pared to other clinical comparison groups, orwhether they may be nonspecific—that is, they maybe similarly elevated in those presenting for treat-ment of other psychiatric disorders.

The literature on the impact of comorbidity onoutcome in AN is mixed. Some studies report noimpact of comorbidity9,10; however, other studiesreport poorer outcome when comorbid disordersare present.11–17 The impact of comorbidity may beon the individual’s general psychosocial function-ing, rather than on the eating disorder directly.18,19

Concurrent comorbidity has been reported to bemore pronounced in the acute phase of AN, withdepression and obsessionality either secondaryto or exacerbated by the starvation state.20,21

Improvement in comorbid symptoms often ac-companies weight restoration, however persisting

Accepted 15 July 2007

1 Department of Psychological Medicine, University of Otago,

Christchurch School of Medicine and Health Sciences,

Christchurch, New Zealand2 Department of Respiratory Medicine, Christchurch Hospital,

Canterbury District Health Board, Christchurch, New Zealand3 Department of Psychiatry, University of North Carolina at

Chapel Hill, North Carolina4 Department of Nutrition, University of North Carolina at

Chapel Hill, North Carolina

*Correspondence to: Jennifer Jordan, Department of Psychological

Medicine, University of Otago, Christchurch School of Medicine and

Health Sciences, PO Box 4345, Christchurch 8140, New Zealand.

E-mail: [email protected]

Supported from Health Research Council of New Zealand.

Published online 14 September 2007 in Wiley InterScience

(www.interscience.wiley.com). DOI: 10.1002/eat.20463

VVC 2007 Wiley Periodicals, Inc.

International Journal of Eating Disorders 41:1 47–56 2008—DOI 10.1002/eat 47

REGULAR ARTICLE

obsessional symptoms and depression have beenfound in patients recovered from AN.22–24 Severalstudies have found reduction in some personalitydisorder symptoms with successful treatment ofeating disorders.25,26 Reductions in personality dis-order symptoms have also have been reported inthose with depression.27,28

Specific elevated prevalences in obsessive com-pulsive disorder (OCD) diagnoses in AN samples,and specific elevations in psychoactive substanceuse disorders (PSUD), and Cluster B diagnoses inbulimia nervosa (BN) (and where data are avail-able, often also in the AN-BP subtype) have beenreported.29–31 Consistent reports also exist of non-specific elevations in anxiety disorders, mood disor-ders, and Cluster C personality diagnoses in eatingdisorder samples relative to controls.30,31

Meta-analyses which have attempted to bringunity to the literature, report wide confidence inter-vals for observed prevalences (e.g., Ref. 31) andinconsistencies and gaps in the literature remaindespite improved methodological practices overthe past 10–15 years, including the increased use ofstandardized assessments. Most unclear is thequestion of comorbidity across AN subtypes, inlarge part due to the relative recency of formal ANsubtyping. One recent meta-analysis cited insuffi-cient studies to report data for the AN-BP sub-type.29

Before important questions of the relative impactof comorbidity on AN course and outcome can beanswered, a clearer picture of comorbid diagnosticprofiles in AN is required. Adequate clarificationwill include the extent to which observed patternsof comorbidity are specific to AN or reflect nonspe-cific elevations in comorbidity as observed acrossclinical samples of all psychiatric diagnoses. Thisquestion can best be addressed by use of clinicalcomparison groups.

The aim of this study is to examine pretreatmentcharacteristics and lifetime prevalences of Axis Iand II comorbid disorders in a sample of womenpresenting for outpatient treatment of AN in com-parison with two other outpatient samples—women with BN and women with major depressivedisorder—to establish whether patterns of comor-bidity seen in AN are disorder-specific or commonacross similarly ascertained samples. We hypothe-size that while high levels of anxiety disorders willbe found in all three samples, specific elevations inprevalences of OCD will be found in the AN sampleand lower levels of PSUDs will be found in the ANsample, relative to the BN sample.

Method

Participants

Participants were drawn from four different random-

ized controlled trials conducted within the Department

of Psychological Medicine in the Christchurch School of

Medicine and Health Sciences. Full details of the assess-

ment and treatment for these three samples are

described elsewhere for the AN sample,32,33 the BN sam-

ple,34 and the depression sample (DEP).35–37 There was

no overlap across these samples in actual participants.

All studies received ethical approval from the Southern

Regional Health Authority Ethics Committee (Canter-

bury) and participants provided written informed

consent.

Anorexia Nervosa Sample

Participants were 56 women aged 17–40 years with

current primary ‘‘spectrum’’ AN, presenting for outpa-

tient psychotherapy for AN.32 Spectrum AN included

those with AN diagnosed according to strict (BMI\ 17.5)

or lenient (BMI 17.5\ 19) weight criteria. For both strict

and lenient AN, the amenorrhoea criterion was not nec-

essary for inclusion in the study, given debate in the liter-

ature as to the necessity of this diagnostic criterion.38

Previous analyses33 reported few differences between

strict and lenient subgroups. The AN sample was split

into subtypes based on a cut-off of at least once weekly

binge or purge episodes over the past 3 months. This def-

inition was based on DSM-IV text suggesting that most

individuals who binge or purge do so at least weekly, and

on empirical grounds—this definition proved to be the

most able to discriminate differences within the AN sam-

ple on a range of variables in previous analyses.39 Exclu-

sion criteria were current severe major depression (DEP)

or serious suicidal intent, current severe psychoactive

substance dependence, cognitive impairment, bipolar I

disorder, schizophrenia, severe physical illness, severe

medical complications of AN or a chronic treatment-re-

fractory course of AN (5 or more years with AN and fail-

ure to respond to two or more adequate courses of treat-

ment). Recruitment was broad-based, and included

referrals from health professionals and self-referral. The

trial was presented at meetings with mental health and

community based health professionals, pamphlets were

sent to health professionals and to community facilities

such as gyms and libraries, and advertisements were

placed in community newspapers.

Bulimia Nervosa Sample

Participants were 132 women between 17 and 40 years

presenting for treatment for BN.34 Exclusion criteria were

current AN or a BMI greater than 30, current severe major

depression or serious suicidal intent, current severe

psychoactive substance dependence, bipolar I or schizo-

JORDAN ET AL.

48 International Journal of Eating Disorders 41:1 47–56 2008—DOI 10.1002/eat

phrenia, current antisocial personality disorder, develop-

mental learning disorder or cognitive impairment, major

medical or neurological illness, current significant medi-

cal complications of BN and current use of psychoactive

medication. Three participants entered into the original

study but were excluded prior to being randomized to

the experimental condition because of deterioration in

mental state requiring further treatment. The reasons for

exclusion were severe depression (n5 2) and severe alco-

hol abuse (n 5 1). Recruitment strategies were the same

as in the AN sample.

Depression Sample

Participants were 100 women with major depressive

disorder recruited in two previously conducted antide-

pressant trials.35,36 Included in this sample were all

women in both studies in the 18–40 age range (to match

the AN sample), but excluded were those with past or

current AN or BN. All met criteria for current major

depressive disorder. Exclusion criteria were bipolar I

disorder, significant medical illness, current severe psy-

choactive substance dependence. Those with suicidal

ideation were not excluded. Recruitment strategies were

the same for both depression studies, consisting primarily

of health professional referrals but self-referrals were also

accepted. Mental health professionals and community

based general practitioners were informed of the studies

in formal presentations, meetings and by word of mouth.

Participants in the DEPs were not recruited by advertising.

Measures

Baseline data included demographic details (marital

status, ethnicity, and age), age of onset and duration of

the index condition, and history of suicide attempts. The

patient version of the Structured Clinical Interview for

DSM-III-R (with psychotic screen) Version 1.0 (SCID-P)

and the Structured Clinical Interview II for DSM-III-R

(SCID II) were administered at baseline by trained clini-

cal interviewers (psychiatrists, senior psychiatric regis-

trars, or clinical psychologists) to determine the presence

of lifetime mood, anxiety, PSUD, and personality disor-

ders. Axis II data were available for 49 of the 56 partici-

pants in the AN sample. The Hamilton Depression Rating

Scale (HDRS, 17-item) is a widely used measure of the se-

verity of depression in patients diagnosed with the disor-

der. The Global Assessment of Functioning (GAF) is Axis

V of the DSM-IV and is designed as a clinician-rated mea-

sure of psychosocial functioning in the last week.

Statistical Analysis

The Statistical Package for the Social Sciences (SPSS,

Version 12) was used to analyze data. Analysis of variance

between samples for independent samples and student

t-tests for AN subtype analyses were used to compare

normally distributed continuous variables. The Kruskal–

Wallis test was used for variables with nonparametric dis-

tributions. Where the ANOVA or Kruskal–Wallis tests indi-

cated significant differences among the diagnostic

groups, these were further explored using independent t-

tests or Mann–Whitney U-tests as appropriate. For di-

chotomous variables, the v2 test was used, with Fisher’s

exact test used where expected cell numbers were small.

To reduce the risk of Type I error we have opted for a

more stringent statistical significance level of\.01 but by

doing this the risk of Type II error is increased. Power cal-

culations however indicated that we had sufficient power

to pick up differences of between 15 and 25% at 80%

power (with p\ .01) where the prevalences of disorder in

the smallest/lowest group were between 3 and 50%.

Binary logistic regression analyses were used to deter-

mine the most parsimonious set of predictors of group

membership for each comparison: AN versus BN, AN ver-

sus DEP, and for AN-R versus AN-BP subtypes. Variables

with a p-value of .05 or smaller in the univariate analyses

were entered into the regressions. Age was a statistically

significant variable in univariate analyses, with the AN

group being younger than both the BN and DEP sam-

ples and this was duly entered into all relevant regres-

sions analyses as a covariate. For the AN subtype analy-

sis, it was necessary to use the number of Cluster B

symptoms rather than the dichotomous ‘any’ Cluster B

diagnosis variable, as all Cluster B diagnoses fell within

the AN-BP subtype, meaning the logistic regression

model could not be fitted using the dichotomous vari-

able.

Inter-rater reliability checks were undertaken by a sec-

ond independent rater blind to the original diagnoses,

rerating greater than 10% of video-taped interviews.

Kappa statistics were calculated to establish inter-rater

reliability for SCID I and II diagnoses in the AN sample.

The overall kappa for lifetime Axis I diagnoses was 0.85

(overall concordance 93%, range 83–100%), and for Axis

II disorders was 0.81 (overall concordance 96%, range 80–

100%). The kappa for any personality disorder diagnosis

in the Depression study was 0.78 (already reported in

Ref. 40). Data were not available for Axis I in the DEP, nor

for the BN sample.

Results

Table 1 presents descriptive characteristics of thethree samples. The AN sample was younger thanboth comparison groups and had an earlier age ofonset of the index condition than the BN sample;however there were no significant differencesacross all three samples in ethnicity, marital status,duration of illness or history of suicide attempts.

SPECIFIC AND NONSPECIFIC COMORBIDITY

International Journal of Eating Disorders 41:1 47–56 2008—DOI 10.1002/eat 49

Ethnicity was predominantly New Zealand Cauca-sian in all studies. The number of participants ofother ethnicities were as follows: Maori (AN n 5 0,BN n 5 8, Depression (DEP) n 5 2), Pacific Island(AN n 5 0, BN n 5 2, DEP n 5 1), Asian (AN n 5 0,BN n 5 1, DEP n 5 3), other ethnicity (AN n 5 2,BN n 5 1, DEP n 5 0). The AN sample had lowerGlobal Assessment of Functioning scores than theBN sample, but did not differ from the DEP. Aswould be expected, the DEP was more depressedthan the eating disorder samples; however, the ANsample had significantly higher current depressionscores (HDRS) than the BN sample.

Table 2 compares specific mood, anxiety, andPSUD prevalences across the three samples. Thosewith AN had significantly higher prevalences ofOCD compared to the BN and DEP samples. The

AN and BN samples had similar levels of depres-sion diagnoses (the DEP were excluded from thisanalysis). However, both the AN and the DEP sam-ples had significantly lower levels of ‘‘any’’ PSUD,and alcohol abuse or dependence relative to the BNsample. There were trends for higher levels of socialand simple phobias in both ED samples versus theDEP sample, and for the BN sample to have higherprevalences of Bipolar II and lower levels of panicand/or agoraphobia versus the other samples; how-ever these results did not reach statistical signifi-cance. There were no significant differences betweensamples for prevalences of ‘‘any’’ anxiety disorder,cannabis, or other PSUD abuse or dependence.

Table 3 compares prevalences of Axis II personal-ity disorder diagnoses across the AN, BN, andDEP samples. The AN and the DEP samples had

TABLE 2. Prevalences for Axis I comorbid diagnoses in the AN, BN, and DEP samples

Lifetime Axis 1 Diagnoses AN n5 56 (%) BN n5 132 (%) DEP n 5 100 (%) Statistic1 p

Any mood disorder 68 71 100 .21 .652

Major depressive disorder 63 51 93 .22 .142

Bipolar II 4a 17b 7a 9.3 .013

Any anxiety disorder 55 50 41 3.4 .18Obsessive compulsive disorder 21a 3b 2b 27.4 \.001Panic dis. and/or agoraphobia 25 11 21 6.5 .04Social phobia 30 30 17 6.1 .05Simple phobia 25 27 13 6.7 .04

Any PSUD4 abuse/dependence 34a 49b 27a 11.7 .003Alcohol abuse/dependence 27a 46b 25a 13.3 .001Cannabis abuse/dependence 21 22 13 3.3 .19Other PSUD abuse/dependence 9 11 4 4.1 .13

1 Statistics are v2.2 Comparisons are between AN and BN (not computed for DEP as presence of depression was an inclusion criterion).3Posthoc analyses for significant differences (p\ 0.01) indicated by letters ‘‘a’’ and ‘‘b’’—groups which share a letter do not differ significantly.4Psychoactive substance use disorder.

TABLE 1. Descriptive characteristics of participants within the anorexia nervosa (AN), bulimia nervosa (BN), anddepression (DEP) samples

AN n 5 56Mean (6 sd) orMedian (Range)

BN n5 132Mean (6 sd) orMedian (Range)

DEP n5 100Mean (6 sd) orMedian (Range) Statistic1 p

DemographicAge (years) 20.5 (16–40)a 25.0 (17–40)b 25.0 (18–40)b 14.1 .0012

EthnicityNZ Caucasian 96% 91% 94% 2.1 .36

Marital statusNever married 63% 62% 58% .5 .78

Psychiatric historyAge of onset (year) 16.0 (8–38)a 18.5 (10–36)b 18.0 (6–37) 14.1 .001Duration (years) 3.5 (0–24) 5.0 (0–25) 5.5 (0–30) 3.1 .21Prior suicide attempt 27% 31% 39% 2.2 .33

Current functioningGAF3 50.0 (38–61)a 55.0 (41–72)b 52.0 (11–65)a 43.2 \.001HDRS4 12.6 (6.9)a 8.6 (5.7)b 19.9 (4.9) 4.1 \.0015

1 Statistics are v2 or Fisher’s exact test for dichotomous variables, and Anovas or Kruskal–Wallis tests for continuous variables.2 Posthoc analyses for significant differences (p\ 0.01) indicated by letters ‘‘a’’ and ‘‘b’’—groups which share a letter do not differ significantly.3Global Assessment of Functioning.4Hamilton depression rating scale (17 item).5 Comparison is between AN and BN (not computed for DEP as presence of depression was an inclusion criterion).

JORDAN ET AL.

50 International Journal of Eating Disorders 41:1 47–56 2008—DOI 10.1002/eat

significantly lower any Cluster B (and borderlineand histrionic) personality disorders than the BNsample. There were trends for significantly higherany Cluster A personality disorder and obsessive-compulsive personality disorder (OCPD) diagnosesin both ED samples compared to the DEP sample.Prevalences of ‘‘any’’ personality disorder and anyCluster C personality disorder diagnoses did notdiffer across the three samples.

Table 4 presents the results of binary logisticregression analyses used to identify the best varia-bles (where p\ .05 on univariate analysis) predict-ing membership of the AN versus BN samples, andthe AN versus DEP samples respectively. For theAN versus BN analysis, four variables remained inthe model with p-values \ .05: higher GlobalAssessment of Functioning scores, Bipolar II, andany Cluster B personality disorder predicted BN,and OCD predicted AN. Using a backwards condi-tional regression analysis, age also entered themodel (predicting BN status) along with the otherfour variables; however this did not reach statisticalsignificance.

For the AN versus the DEP comparison (seeTable 4), robust main effects emerged for age (olderage predicting DEP) and OCD (predicting AN)using both forwards and backwards conditionalregression analyses.

Table 5 presents Axis I comorbidity data for ANsubtypes. For Axis I disorders, there were no statis-tically significant differences at an alpha level ofp\ .01. There were however trends for the AN-BPsubtype to have higher prevalences of any PSUDand cannabis abuse or dependence compared tothe AN-R subtype. Levels of any PSUD for the AN-

BP subtype (48%) were at a similar level to thatseen in the BN group (49%). In addition, we exam-ined whether the subtypes differed by age. Therewas no significant difference for age by AN subtype.Medians and ranges were AN-R 23 years (17–40years) and AN-BP 19 years (16–40 years), Mann–Whitney U 336.50, p 5 .40).

Table 6 compares Axis II comorbidity within ANsubtypes. The AN-BP subtype had significantlyhigher prevalences of any Cluster B personality dis-order diagnosis. There were trends approachingsignificance for higher prevalences in AN-BP sub-

TABLE 3. Axis II comorbidity within the AN, BN, and DEP samples

Lifetime Axis II Diagnoses AN n5 49 (%) BN n5 132 (%) DEP n5 96 (%) Statistic1 p

Any personality disorder (PD)2 51 57 46 2.7 .26Cluster A personality disorder 25 28 13 8.0 .02Cluster B personality disorder 12a 34b 18a 12.9 .0023

Cluster C personality disorder4 37 42 33 2.0 .37Paranoid PD 20 24 13 4.9 .09Schizoid PD 6 2 1 .11Schizotypal PD 2 5 0 4.8 .09Antisocial PD 2 5 2 .35Borderline PD 10a 29b 16a 10.2 .006Histrionic PD 4a 17b 5a 10.4 .006Narcissistic PD 6 5 0 .07Avoidant PD 31 29 23 1.4 .51Dependent PD 18 13 10 1.8 .40Obsessive-compulsive PD 18 18 6 7.5 .02

1Statistics are v2 or Fishers Exact Test for dichotomous measures.2 Axis II data were only available for 49/56 participants in the AN sample.3Posthoc analyses for significant differences (p\ 0.01) indicated by letters ‘‘a’’ and ‘‘b’’—groups which share a letter do not differ significantly.4 These variables are based on the DSM-IV composition of Cluster C personality disorder (three diagnoses only-Avoidant, Dependent, and Obsessive-com-

pulsive).

TABLE 4. Results of binary logistic regression analysesbetween the AN and BN samples, and the AN andDEP samples

VariableOddsRatio

95% ConfidenceInterval p

AN versus BN1

Forward conditionalGlobal assessment of functioning 1.23 1.15–1.37 \.001Bipolar II 4.00 1.32–12.13 .01Obsessive compulsive disorder 0.42 0.21–0.85 .02Any cluster B personality disorder 2.81 1.48–5.36 .002

Backwards conditionalAge 1.07 0.99–1.16 .08Global assessment of functioning 1.24 1.14–1.36 \.001Bipolar II 3.79 1.25–11.50 .02Obsessive compulsive disorder 0.36 0.17–0.76 .008Any cluster B personality disorder 2.90 1.50–5.61 .002

AN versus DEP2

Age 1.12 1.05–1.19 .001Obsessive compulsive disorder 0.22 0.09–0.50 \.001

1Variables entered into the model for AN versus BN: Age, GAF, HDRS,OCD, Panic, and/or agoraphobia, Bipolar II, any PSUD diagnosis, any clus-ter B diagnosis. Variables were entered if p � 0.05 on univariate analysis.

2 Variables entered into the model for AN versus DEP: Age, OCD, Socialphobia, any cluster A diagnosis, OCPD. Variables were entered if p �0.05on univariate analysis.

SPECIFIC AND NONSPECIFIC COMORBIDITY

International Journal of Eating Disorders 41:1 47–56 2008—DOI 10.1002/eat 51

type of any Cluster A, paranoid and borderline per-sonality disorder diagnoses. Once again, the AN-BPlevels resembled the BN sample. The AN-R grouphad no Cluster B diagnoses.

Binary logistic regression analysis was used toidentify the best variables to predict membershipof the AN-R versus AN-BP groups. These results arepresented in Table 7. Using a forward conditionalprocedure, paranoid personality disorder remainedin the model at a significance level of p\ .05, pre-dicting the AN-BP subtype. Using a backward step-wise conditional model, although paranoid person-ality disorder and number of Cluster B symptoms(the dimensional variable substituted for the any

Cluster B diagnosis variable) remained in themodel predicting the AN-BP subtype, these weretrends and did not reach significance.

Conclusion

We compared comorbid Axis I and II diagnoses inan AN outpatient clinical sample relative to sam-ples of individuals with BN or major depressive dis-order to establish which comorbid associationswere specific to AN versus nonspecific elevationscommon to other clinical samples.

Comorbidity was common in these three clinicalsamples. However, there was evidence of specificassociations with AN (OCD), specific elevations inBN (PSUD, Cluster B diagnoses), and nonspecificelevations across samples (anxiety and Cluster Cpersonality disorders). There were also trends forspecific associations with eating disorders (socialand simple phobias, Cluster A personality disordersand OCPD) relative to the DEP sample. The AN-BPsubtype resembled BN for some PSUD, and ClusterB personality disorder diagnoses.

The elevated levels and patterns of comorbidityseen here in the AN, BN, and DEP samples arelargely consistent with that reported in the clinicalliterature for eating disorders30,31 and for majordepression.41 Previous studies have also reportedspecific elevations of OCD in AN42–45 (althoughcontrary to one recent study46), lower PSUD,47 andCluster B diagnoses in AN relative to those withBN,29,48 and in those with AN-R compared to thosewith AN-BP.30,49,50 Other studies have also reportedfindings of elevated but nonspecific comorbidity formood,2,24,47,51–54 anxiety,24,46,55 and Cluster C per-sonality disorder.29

Our findings are in contrast to studies reportinglower prevalences of lifetime major depression inthose with the AN-R subtype relative to the AN-BP

TABLE 6. Axis II comorbidity in the anorexianervosa subtypes

Lifetime Axis II Diagnoses

AN Subtypes

Statistic1 PAN-R

n5 26 (%)AN-BP

n5 232 (%)

Any personality disorder (PD)1 39 65 3.5 .06Any cluster A personality disorder 12 39 5.0 .03Any cluster B personality disorder 0 26 .007Any cluster C personality disorder3 27 48 2.3 .13Paranoid PD 8 35 5.5 .02Schizoid PD 4 9 .59Schizotypal PD 0 4 .47Antisocial PD 0 4 .47Borderline PD 0 22 .02Histrionic PD 0 9 .22Narcissistic PD 0 13 .10Avoidant PD 19 44 3.4 .07Dependent PD 12 26 .27Obsessive-compulsive PD 12 26 .27

1Statistics are v2 or Fishers Exact Test.2 Axis II data were available for 49 of the 56 participants.3 This variable is based on the DSM-IV composition of Cluster C personal-

ity disorders (three diagnoses only-Avoidant, Dependent and Obsessive-Compulsive).

TABLE 5. Axis I comorbidity in anorexia nervosasubtypes

Lifetime Axis I Diagnoses

AN Subtypes

Statistic1 pAN-R

n 5 31 (%)AN-BP

n5 25 (%)

Any mood 68 68 .0 .98Major depressive disorder 58 68 .6 .45Bipolar II 7 0 .50

Any anxiety disorder 55 56 .01 .93Obsessive compulsive disorder 19 24 .2 .67Panic dis. and/or agoraphobia 23 28 .2 .64Social phobia 29 32 .0 .81Simple phobia 23 28 .2 .64

Any PSUD2 abuse/dependence 23 48 4.0 .05Alcohol abuse/dependence 23 32 .6 .43Cannabis abuse/dependence 10 36 5.7 .02Other PSUD Abuse/dependence 7 12 .65

1Statistics are v2 or Fishers Exact Test for dichotomous measures.2 Psychoactive Substance Use Disorder.

TABLE 7. Results of binary logistic regression analysisbetween the AN-R and AN-BP subtypes

VariableOddsRatio

95% ConfidenceInterval P

AN-R versus AN-BP1

Forward conditionalParanoid personality disorder 2.53 1.09–5.86 .03

Backwards conditionalParanoid personality disorder 2.23 0.94–5.31 .07Number of cluster B personality

disorder symptoms1.14 0.98–1.33 .08

1Variables entered into the model for AN-R versus AN-BP: Any PSUD,number of Cluster B symptoms, Paranoid Personality Disorder. Variableswere entered into the model if p � 0.05 on univariate analyses.

JORDAN ET AL.

52 International Journal of Eating Disorders 41:1 47–56 2008—DOI 10.1002/eat

subtype.47,55,56 Furthermore, unlike Bulik et al.,57

who reported significantly higher alcohol use disor-ders in those with histories of both AN and BN, andthe AN-BP subtype compared to those with AN-R,in the present study the differences between ANsubtypes emerged for cannabis rather than alcoholabuse or dependence, although this was only at atrend level. Another difference was the relativelyhigh level of alcohol and cannabis abuse or de-pendence in the AN-R subtype compared to previ-ous AN samples.11,57,58 The AN-R PSUD levels inthis study were consistent with that in the DEPsample, with samples with other psychiatric condi-tions (e.g. Ref. 59) and indeed within youngercohorts in New Zealand community samples.60,61

Of note was the high ‘‘any’’ Cluster A personalitydisorder prevalence in both eating disorder sam-ples, second in frequency to Cluster C in the ANsample. Cluster A personality disorders are infre-quently reported in eating disorder samples—someauthors have stated that, ‘‘as expected, virtually nopatients received a Cluster A diagnosis’’ (Ref. 62, p578). Where Cluster A diagnoses are reported, moststudies report low rates but with paranoid person-ality disorder being the most common in this clus-ter.29,63 Previous reports about this BN sampleappear to be the only reports of such high rates inBN64,65 and there are no other reports of frequen-cies of this magnitude for AN.29

Lilenfeld et al.’s review of the evidence for modelsof personality comorbidity in eating disorders con-cludes that obsessional traits are likely to be riskfactors across AN and BN, but especially for theAN-R subtype.66 In addition, underlying impulsivetraits have also been implicated in the associationof BN (and the AN-BP subtype) with PSUD andCluster B personality disorders.18,29,57

Results from the present study suggest that it ispremature to narrow our investigations of person-ality pathology in eating disorders to one or twospecific disorders of interest. Borderline personalitydisorder has received the most attention in studiesof BN, however Cluster C and avoidant personalitydisorder are also relevant in terms of functionalinterrelationships with eating disorders. This rela-tive neglect may be because there tends to be ageneral elevation in Cluster C disorders acrossother clinical samples, as seen here in the DEPsample. These findings also emphasize the impor-tance of continuing to collect and report data byAN subtypes.

A strength of the present study was the com-parison of comorbidity using structured clinicalinterviews in three outpatient clinical samples with

similar demographics. It also provided a completelist of all Axis II diagnoses in AN by subtype and incomparison to a BN sample and a clinical compari-son group, allowing examination of specific versusnonspecific comorbidity.

The present study has a number of limitations.These are clinical samples and so may be biased byknown5,6 and unknown effects of sample selection.Although there was overlap in recruitment strat-egies in seeking health professional referral andaccepting self-referral, both ED studies used morebroad based community recruitment strategiesincluding advertising. The likely effect of this wouldbe a bias towards less illness severity and comor-bidity in the ED samples relative to the DEP sam-ples. However with the exception of the small butstatistically significant finding for higher GAFscores in the BN sample, these data suggest thatthe ED samples were no less severe in psychiatrichistory variables and the direction of the comorbid-ity findings was for equivalent or higher prevalen-ces in the ED samples than in the DEP sample.

The AN sample is relatively small, especially forsubtype analyses and for examining low base ratecomorbid disorders, raising the risk of insufficientpower leading to Type II error–specifically, the fail-ure to detect specific differences in comorbiditywhich may actually be there. We acknowledge thatwe have limited power to detect differencesbetween the groups for low base rate diagnoses. Inaddition, by attending to Type I error to allow tosome degree for the number of comparisons, thishas also reduced our power for some comparisons.However, low base rate diagnoses may be still clini-cally relevant and for some comparisons (e.g., forAN subtypes), these data may not have been pre-sented previously.

The exclusion criteria of current severe PSUD(and current severe depression in both eating dis-order clinical trials) pose potential problems in astudy examining comorbidity. It is important toacknowledge that the samples drawn together forthis study examining comorbidity were originallyfrom treatment seeking samples. As such, theexclusion criteria used were stricter than if thestudy had been designed a priori as a comorbiditystudy. Although few were excluded on the basis ofsevere comorbidity, there may be other participantswho were never referred because referrers wereaware of exclusion criteria. The effect of this bias inascertainment would be to under-report the trueprevalence in the pool of potential participants.

However it is important to note that these dataare for lifetime comorbidity. In practice no-one was

SPECIFIC AND NONSPECIFIC COMORBIDITY

International Journal of Eating Disorders 41:1 47–56 2008—DOI 10.1002/eat 53

excluded exclusively on comorbidity grounds in theAN sample, observed prevalence of comorbid life-time disorders were commensurate with thosereported in other studies, and the comorbid pat-terns were similar to those seen in epidemiologicalstudies. Although three participants had beenexcluded from the BN study because of deteriora-tion in comorbid conditions, as for the AN sample,levels and patterns for comorbid depression andPSUD are similar to that seen in other studies ofthose with eating disorders. Suicidal ideation wasnot an exclusion criterion in the DEP sample butserious suicidal ideation was within both ED stud-ies; however the difference between all three sam-ples for a history of suicide attempts was not statis-tically significant.

Baseline data were collected during the acutephase (presentation for outpatient treatment),which has the potential problem of confoundingAxis I and II diagnoses by the distress associatedwith the index condition. This may be particularlypertinent in AN in which starvation may influencemood, increase levels of obsessionality,67 and pos-sibly paranoid tendencies.63

The younger median age of the AN sample mayhave influenced the lower frequencies of some dis-orders such as panic disorder and PSUD whichhave a relatively later age of onset relative toAN.42,51 Individuals in the AN sample may not havepassed through the age of risk for onset of thesedisorders to the same extent as those in the BN andDEP samples. Of note though, for both panic and/or agoraphobia and PSUD, prevalences in theyounger AN sample did not differ from the olderDEP comparison sample, suggesting that factorsother than age might account for the differencesbetween the AN and BN samples for those varia-bles. In addition, although age remained significantin the final logistic regression analyses between ANand the other two samples, it is important to notethat the other comorbidity variables remained sig-nificant despite age having a main effect.

Although the inter-rater reliability figures whereavailable were acceptable, the absence of data forthe BN sample and Axis I for the DEPs is a limita-tion. However, there was some overlap in investiga-tors and raters between these studies, with use ofstandardized SCID I and II assessments. In addi-tion, apart from the high Cluster A diagnoses inboth eating disorder studies, the comorbidity prev-alences obtained here are largely consistent withprevalences reported in the literature.

Future research is recommended in the followingareas. Replications of the high levels of Cluster A

personality disorder comorbidity in the AN and BNsamples is required. The relationships among Clus-ter A personality disorder, avoidant personality dis-order, and social phobia, especially in relation toBN and AN-BP diagnoses also warrants further ex-ploration. Alongside the use of categorical diagno-ses for comorbid conditions, the use of continuoussymptom measures where possible will increasepower to detect differences in the context of rela-tively small AN samples (especially for AN sub-types). Finally, rather than excluding individualsfrom clinical trials on the basis of current comor-bidity, it would be valuable to include these indi-viduals but re-assess Axis I and II comorbidity aftertreatment to determine the extent to which theactive illness (e.g., starvation in anorexia nervosa,uncontrolled binge eating and purging in BN) mayinfluence symptom presentation of comorbid dis-orders.

To conclude, although comorbidity was commonin these three clinical samples, there was evidenceof specific associations with AN (OCD), specificassociations with BN (PSUD, Cluster B diagnoses),and nonspecific elevations across samples (anxietyand Cluster C personality disorders). There werealso trends for specific associations with eating dis-orders (social and simple phobias, Cluster A per-sonality disorders and OCPD) relative to the DEPcomparison sample. The AN-BP subtype resembledBN for some PSUD, and Cluster B personality dis-order diagnoses. These results contribute to thesparse AN literature for AN subtypes and Axis IIdisorders. The present study lends further supportto the subtyping of AN, as in terms of comorbidity,those with AN-BP appear to sit between those withAN-R and those with BN, with aspects of overlapwith both AN and BN.

Although consideration of the current DSM sub-typing categorization is beyond the scope of this ar-ticle, there is merit in continuing to consider theAN-BP subtype as a distinct group. This is war-ranted because the difference in comorbidity pro-file as demonstrated here, the presence of bingeingand purging behaviors, and the incumbent seriousphysical risks accompanying both AN and BN areall likely to have a marked influence on treatmentdelivery. The older DSM-III-R classification ofassigning both AN and BN diagnoses may have bet-ter represented this subtype conceptually; howeverthe high bingeing threshold would have excludedmany who are now able to be described within theAN-BP subtype.

This study confirms the presence of specificcomorbidity in AN relative to other clinical groups,

JORDAN ET AL.

54 International Journal of Eating Disorders 41:1 47–56 2008—DOI 10.1002/eat

and between AN subtypes, an important findingwhich has obvious etiological implications. How-ever, nonspecific comorbidity does not necessarilyimply that comorbidity common to other clinicaldiagnostic groups is unimportant. Further atten-tion is needed to examine the relative impact ofspecific versus nonspecific comorbidity on morbid-ity and outcome in AN.

The authors thank Andrea Bartram, Leslie Livingston,Isobel Stevens, Gill Ebel, Barbara Malthus, and RobynAbbott for their assistance in the coordination of thesestudies and Georgina Brooks in the preparation of thismanuscript.

References

1. Bulik CM, Sullivan PF, Carter FA, Joyce PR. Lifetime comorbidity

of alcohol dependence in women with bulimia nervosa. Addict

Behav 1997;22:437–446.

2. Duncan AE, Neuman RJ, Kramer J, Kuperman S, Hesselbrock V,

Reich T, et al. Are there subgroups of bulimia nervosa based on

comorbid psychiatric disorders? Int J Eat Disord 2005;37:19–25.

3. Matsunaga H, Kiriike N, Miyata A, Iwasaki Y, Matsui T, Fujimoto

K, et al. Prevalence and symptomatology of comorbid obses-

sive-compulsive disorder among bulimic patients. Psychiatry

Clin Neurosci 1999;53:661–666.

4. Kessler RC, Berglund P, Demler O, Jin R, Merikangas KR, Walters

EE. Lifetime prevalence and age-of-onset distributions of DSM-

IV disorders in the national comorbidity survey replication.

Arch Gen Psychiatry 2005;62:593–602.

5. Wells JE, Bushnell JA, Hornblow AR, Joyce PR, Oakley-Browne

MA. Christchurch psychiatric epidemiology study, Part 1: Meth-

odology and lifetime prevalence for specific psychiatric disor-

ders. Aust NZ J Psychiatry 1989;23:315–326.

6. Berkson J. Limitations of the application of fourfold table anal-

ysis to hospital data. Biometrics 1946;2:47–53.

7. Bushnell JA, Wells JE, McKenzie JM, Hornblow AR, Oakley-

Browne MA, Joyce PR. Bulimia comorbidity in the general pop-

ulation and in the clinic. Psychol Med 1994;24:605–611.

8. Fairburn CG, Welch SL, Norman PA, O’Connor ME, Doll HA. Bias

and bulimia nervosa: How typical are clinic cases? Am J Psychi-

atry 1996;153:386–391.

9. Halmi KA, Agras WS, Crow S, Mitchell J, Wilson GT, Bryson SW,

et al. Predictors of treatment acceptance and completion in an-

orexia nervosa—Implications for future study designs. Arch

Gen Psychiatry 2005;62:776–781.

10. North C, Gowers S. Anorexia nervosa, psychopathology, and

outcome. Int J Eat Disord 1999;26:386–391.

11. Herpertz-Dahlmann B, Muller B, Herpertz S, Heussen N, Hebe-

brabd J, Remschmidt H. Prospective 10-year follow-up in ado-

lescent anorexia nervosa—Course, outcome, psychiatric comor-

bidity, and psychosocial adaptation. J Child Psychol Psychiatry

2001;42:603–612.

12. Hsu G, Crisp A, Callender J. Psychiatric diagnoses in recovered

and unrecovered anorectics 22 years after onset of illness: A

pilot study. Compr Psychiatry 1992;33:123–127.

13. Lowe B, Zipfel S, Buchholz C, Dupont Y, Reas DL, Herzog W.

Long-term outcome of anorexia nervosa in a prospective 21-

year follow-up study. Psychol Med 2001;31:881–890.

14. Rossiter E, Agras W, Telch C, Schneider J. Cluster B personality

disorder characteristics predict outcome in the treatment of

bulimia nervosa. Int J Eat Disord 1993;13:349–357.

15. Saccomani L, Savoini M, Cirrincione M, Vercellino F, Ravera G.

Long-term outcome of children and adolescents with anorexia

nervosa: Study of comorbidity. J Psychosom Res 1998;44:565–

571.

16. Wentz E, Gillberg C, Gillberg IC, Rastam M. Ten-year follow-up

of adolescent-onset anorexia nervosa: Psychiatric disorders

and overall functioning scales. J Child Psychol Psychiatry 2001;

42:613–622.

17. Wonderlich S, Mitchell JE. The role of personality in the onset

of eating disorders and treatment implications. Psychiatr Clin

North Am 2001;24:249–258.

18. Bruce KR, Steiger H. Treatment implications of Axis-II comor-

bidity in eating disorders eating disorders. 2005;13:93–108.

19. Grilo CM. Recent research of relationships among eating disor-

ders and personality disorders. Curr Psychiatry Rep 2002;4:18–

24.

20. Keys A, Brozek J, Honschel A, Mickelson O, Taylor HL. The Biol-

ogy of Human Starvation. Minneapolis: University of Minnesota

Press, 1950.

21. Vitousek K, Manke F. Personality variables and disorders in an-

orexia nervosa and bulimia nervosa. J Abnorm Psychol

1994;103:137–147.

22. Pollice C, Kaye WH, Greeno CG, Weltzin TE. Relationship of

depression, anxiety, and obsessionality to state of illness in an-

orexia nervosa. Int J Eat Disord 1997;21:367–376.

23. Klump KL, Strober M, Bulik CM, Thornton L, Johnson C, Devlin

B, et al. Personality characteristics of women before and after

recovery from an eating disorder. Psychol Med 2004;34:1407–

1418.

24. Sullivan PF, Bulik CM, Fear JL, Pickering A. Outcome of anorexia

nervosa: A case-control study. Am J Psychiatry 1998;155:939–

946.

25. Ro O, Martinsen EW, Hoffart A, Sexton H, Rosenvinge JH. The

interaction of personality disorders and eating disorders: A

two-year prospective study of patients with longstanding eating

disorders. Int J Eat Disord 2005;38:106–111.

26. Garner DM, Olmstead MP, Davis R, Rockert W, Goldbloom D,

Eagle M. The association between bulimic symptoms and

reported psychopathology. Int J Eat Disord 1990;9:1–15.

27. Mulder RT. Personality pathology and treatment outcome in

major depression: A review. Am J Psychiatry 2002;159:359–371.

28. van den Hout M, Brouwers C, Oomen J. Clinically diagnosed

Axis II co-morbidity and the short term outcome of CBT for Axis

I disorders. Clin Psychol Psychother 2006;13:56–63.

29. Cassin SE, von Ranson KM. Personality and eating disorders: A

decade in review. Clin Psychol Rev 2005;25:895–916.

30. O’Brien KM, Vincent NK. Psychiatric comorbidity in anorexia

and bulimia nervosa: Nature, prevalence, and causal relation-

ships. Clin Psychol Rev 2003;23:57–74.

31. Rosenvinge JH, Martinussen M, Ostensen E. The comorbidity of

eating disorders and personality disorders: A meta-analytic

review of studies published between 1983 and 1998. Eat Weight

Disord 2000;5:52–61.

32. McIntosh VV, Jordan J, Carter FA, Luty SE, McKenzie JM, Bulik

CM, et al. Three psychotherapies for anorexia nervosa: A

randomized controlled trial. Am J Psychiatry 2005;162:741–747.

33. McIntosh VV, Jordan J, Carter FA, McKenzie JM, Luty SE, Bulik

CM, et al. Strict versus lenient weight criterion in anorexia nerv-

osa. Eur Eat Disord Rev 2004;12:51–60.

34. Bulik CM, Sullivan PF, Carter FA, McIntosh VV, Joyce PR. The

role of exposure with response prevention in the cognitive-be-

havioral therapy for bulimia nervosa. Psychol Med 1998;28:

611–623.

SPECIFIC AND NONSPECIFIC COMORBIDITY

International Journal of Eating Disorders 41:1 47–56 2008—DOI 10.1002/eat 55

35. Joyce PR, Mulder RT, Cloninger CR. Temperament predicts clo-

mipramine and desipramine response in major depression.

J Affect Disord 1994;30:35–46.

36. Joyce PR, Mulder RT, Luty SE, Sullivan PF, McKenzie JM, Abbott

RM, et al. Patterns and predictors of remission, response and

recovery in major depression treated with fluoxetine or nortrip-

tyline. Aust NZ J Psychiatry 2002;36:384–391.

37. Mulder RT, Joyce PR, Luty SE. The relationship of personality

disorders to treatment outcome in depressed outpatients.

J Clin Psychiatry 2003;64:259–264.

38. Cachelin F, Maher B. Is amenorrhea a critical criterion for ano-

rexia nervosa? J Psychosom Res 1998;44:435–440.

39. Jordan J. Psychiatric Comorbidity and Subtyping in Anorexia

Nervosa. Christchurch: University of Otago, 2004.

40. Carter JD, Joyce PR, Mulder RT, Sullivan PF, Luty SE. Gender dif-

ferences in the frequency of personality disorders in depressed

outpatients. J Pers Disord 1999;13:67–74.

41. Vuorilehto M, Melartin T, Isometsa E. Depressive disorders in

primary care: Recurrent, chronic, and co-morbid. Psychol Med

2005;35:673–682.

42. Bulik CM, Sullivan PF, Fear J, Joyce PR. Eating disorders and an-

tecedent anxiety disorders: A controlled study. Acta Psychiatr

Scand 1997;96:101–107.

43. Godart NT, Flament MF, Curt F, Perdereau F, Lang F, Venisse JL,

et al. Anxiety disorders in subjects seeking treatment for eating

disorders: A DSM-IV controlled study. Psychiatry Res 2003;117:

245–258.

44. Halmi KA, Eckert E, Marchi P, Sampugnaro V, Apple R, Cohen J.

Comorbidity of psychiatric diagnoses in anorexia nervosa. Arch

Gen Psychiatry 1991;48:712–718.

45. Thornton C, Russell J. Obsessive compulsive comorbidity in the

dieting disorders. Int J Eat Disord 1997;21:83–87.

46. Kaye WH, Bulik CM, Thornton L, Barbarich N, Masters K. Comor-

bidity of anxiety disorders with anorexia and bulimia nervosa.

Am J Psychiatry 2004;161:2215–2221.

47. Braun D, Sunday S, Halmi K. Psychiatric comorbidity in patients

with eating disorders. Psychol Med 1994;24:859–867.

48. Matsunaga H, Kaye WH, McConaha C, Plotnicov K, Pollice C,

Rao R. Personality disorders among subjects recovered from

eating disorders. Int J Eat Disord 2000;27:353–357.

49. Holderness CC, Brooks GJ, Warren MP. Co-morbidity of eating

disorders and substance abuse review of the literature. Int J Eat

Disord 1994;16:1–34.

50. Wonderlich SA, Mitchell JE. Eating disorders and comorbidity:

Empirical, conceptual, and clinical implications. Psychophar-

macol Bull 1997;33:381–390.

51. Deep AL, Nagy LM, Weltzen TE, Radhika R, Kaye WH. Premorbid

onset of psychopathology in long-term recovered anorexia

nervosa. Int J Eat Disord 1995;17:291–297.

52. Halmi KA. Obsessive-compulsive personality disorder and eat-

ing disorders. Eat Disord 2005;13:85–92.

53. Milos GF, Spindler AM, Buddeberg C, Crameri A. Axes I and II

comorbidity and treatment experiences in eating disorder sub-

jects. Psychother Psychosom 2003;72:276–285.

54. Rastam M. Anorexia nervosa in 51 Swedish adolescents: Pre-

morbid problems and comorbidity. J Am Acad Child Adolesc

Psychiatry 1992;31:819–829.

55. Tozzi F, Thornton LM, Klump KL, Fichter MM, Halmi KA,

Kaplan AS, et al. Symptom fluctuation in eating disorders:

Correlates of diagnostic crossover. Am J Psychiatry 2005;162:

732–740.

56. Fornari V, Kaplan M, Sandberg D, Mathews M, Skolnick N, Katz

J. Depressive and anxiety disorders in anorexia nervosa and

bulimia nervosa. Int J Eat Disord 1992;12:21–29.

57. Bulik CM, Klump KL, Thornton L, Kaplan AS, Devlin B, Fichter

MM, et al. Alcohol use disorder comorbidity in eating disorders:

A multicenter study. J Clin Psychiatry 2004;65:1000–1006.

58. Fairburn CG, Cooper Z, Doll HA, Welch SL. Risk factors for ano-

rexia nervosa: Three integrated case-control comparisons. Arch

Gen Psychiatry 1999;56:468–476.

59. Grilo CM, Levy KN, Becker DF, Edell WS, McGlashan TH.

Comorbidity of DSM-III-R Axis I and II disorders among female

inpatients with eating disorders. Psychiatr Serv 1996;47:426–

429.

60. Fergusson DM, Horwood LJ, Ridder EM. Conduct and atten-

tional problems in childhood and adolescence and later

substance use, abuse and dependence: Results of a 25 year lon-

gitudinal study. Drug Alcohol Depend 2007;88(Suppl 1):S14–

S26.

61. Poulton R, Moffitt T, Harrington H, Milne B, Caspi A. Persistence

and perceived consequences of cannabis use and dependence

among young adults: Implications for policy. NZ Med J 2001;

114:544–547.

62. Thompson-Brenner H, Westen D. Personality subtypes in eating

disorders: Validation of a classification in a naturalistic sample.

Br J Psychiatry 2005;186:516–524.

63. Ro O, Martinsen EW, Hoffart A, Rosenvinge J. Two-year prospec-

tive study of personality disorders in adults with longstanding

eating disorders. Int J Eat Disord 2005;37:112–118.

64. Bulik CM, Sullivan PF, Carter FA, Joyce PR. Temperament, char-

acter, and personality disorder in bulimia nervosa. J Nerv Ment

Dis 1995;183:593–598.

65. Mulder RT, Joyce PR, Sullivan PF, Bulik CM, Carter FA. The rela-

tionship among three models of personality psychopathology:

DSM-III-R personality disorder, TCI scores and DSQ defences.

Psychol Med 1999;29:943–951.

66. Lilenfeld LRR, Wonderlich S, Riso LP, Crosby R, Mitchell J. Eating

disorders and personality: A methodological and empirical

review. Clin Psychol Rev 2006;26:299–320.

67. Vitousek KM, Stumpf RE. Difficulties in the assessment of per-

sonality traits and disorders in eating-disordered individuals.

Eat Disord 2005;13:37–60.

JORDAN ET AL.

56 International Journal of Eating Disorders 41:1 47–56 2008—DOI 10.1002/eat