do adolescents with eating disorder not otherwise specified or full-syndrome bulimia nervosa differ...

7
Do Adolescents with Eating Disorder Not Otherwise Specified or Full-Syndrome Bulimia Nervosa Differ in Clinical Severity, Comorbidity, Risk Factors, Treatment Outcome or Cost? Ulrike Schmidt, MD, PhD 1 * Sally Lee, PhD 2 Sarah Perkins, DClinPsy 1 Ivan Eisler, PhD 1 Janet Treasure, MD, PhD 1 Jeny Beecham, PhD 3 Mark Berelowitz, MD 4 Liz Dodge, MSc 5 Susie Frost, DClinPsy 1 Mari Jenkins, MD 5 Eric Johnson-Sabine, MD 5 Saskia Keville, DClinPsy 5 Rebecca Murphy, DClinPsy 1 Paul Robinson, MD 6 Suzanne Winn, DClinPsy 1 Irene Yi, MD 7 ABSTRACT Objective: We wanted to know whether adolescents with eating disorder not oth- erwise specified (EDNOS) differ from those with bulimia nervosa (BN) in clini- cal features, comorbidity, risk factors, treatment outcome or cost. Method: Adolescents with EDNOS (n 5 24) or BN (n 5 61) took part in a trial of family therapy versus guided self-care. At baseline, eating disorder symptoms, risk factors, and costs were assessed by interview. Patients were reinterviewed at 6 and 12 months. Results: Compared with EDNOS, BN patients binged, vomited and purged sig- nificantly more, and were more preoccu- pied with food. Those with EDNOS had more depression and had more current and childhood obsessive-compulsive dis- order. 66.6% of EDNOS versus 27.8% of BN patients were abstinent from binge- ing and vomiting at 1 year. Diagnosis did not moderate treatment outcome. Costs did not differ between groups. Conclusion: EDNOS in adolescents is not trival. It has milder eating disorder symptoms but more comorbidity than BN. V V C 2008 by Wiley Periodicals, Inc. Keywords: bulimia nervosa; ado- lescence; family therapy; EDNOS; guided selfcare (Int J Eat Disord 2008; 41:498–504) Introduction Three to five percent of young women suffer from full or partial-syndrome bulimia nervosa (BN). 1 Typically, the disorder develops in mid-to late adolescence and adolescents now form the majority of those who present for treatment in primary care. 2 There has recently been considerable interest in those eating disorders that do not fulfill precise diagnostic criteria, i.e., those currently classified as eating disorders not otherwise specified (EDNOS 3 ). These cases are the most common groups identi- fied in community samples 4–6 and in out-patient settings. 7 Studies delineating different sub-types (such as purging disorder 8,9 ), risk factors, comor- bidity, course, and outcome are only just beginning to emerge. 8–11 It has been said that many of these cases are clin- ically as severely unwell as those with full syndro- mal eating disorders, 12 and an example presented in support of this statement concerns patients who are of normal weight, binge infrequently, but purge frequently, which is medically dangerous. However, an alternative possibility is that many patients with EDNOS may have milder eating disorders, but are referred because of greater levels Supported by Health Foundation, UK. *Correspondence to: Ulrike Schmidt, MD, PhD, Section of Eating Disorders, Institute of Psychiatry, King’s College, London, UK. E-mail: [email protected] Accepted 20 February 2008 1 Section of Eating Disorders, Institute of Psychiatry, King’s College, London, United Kingdom 2 Clinical Trials Unit, Institute of Psychiatry, King’s College, London, United Kingdom 3 Center for the Economics of Mental Health, Institute of Psychiatry, King’s College, London, United Kingdom 4 Department of Child Psychiatry, Royal Free Hospital, London, United Kingdom 5 The Phoenix Wing Eating Disorders Service, St Ann’s Hospital, London, United Kingdom 6 Russell Unit Eating Disorders Service, Royal Free Hospital, London, United Kingdom 7 Child and Adolescent Eating Disorders Services, Frimley Children’s Center, Frimley, United Kingdom Published online 23 April 2008 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/eat.20533 V V C 2008 Wiley Periodicals, Inc. 498 International Journal of Eating Disorders 41:6 498–504 2008 REGULAR ARTICLE

Upload: ulrike-schmidt

Post on 12-Jun-2016

216 views

Category:

Documents


4 download

TRANSCRIPT

Page 1: Do adolescents with eating disorder not otherwise specified or full-syndrome bulimia nervosa differ in clinical severity, comorbidity, risk factors, treatment outcome or cost?

Do Adolescents with Eating Disorder Not OtherwiseSpecified or Full-Syndrome Bulimia Nervosa Differ inClinical Severity, Comorbidity, Risk Factors, Treatment

Outcome or Cost?

Ulrike Schmidt, MD, PhD1*Sally Lee, PhD2

Sarah Perkins, DClinPsy1

Ivan Eisler, PhD1

Janet Treasure, MD, PhD1

Jeny Beecham, PhD3

Mark Berelowitz, MD4

Liz Dodge, MSc5

Susie Frost, DClinPsy1

Mari Jenkins, MD5

Eric Johnson-Sabine, MD5

Saskia Keville, DClinPsy5

Rebecca Murphy, DClinPsy1

Paul Robinson, MD6

Suzanne Winn, DClinPsy1

Irene Yi, MD7

ABSTRACT

Objective: We wanted to know whether

adolescents with eating disorder not oth-

erwise specified (EDNOS) differ from

those with bulimia nervosa (BN) in clini-

cal features, comorbidity, risk factors,

treatment outcome or cost.

Method: Adolescents with EDNOS (n

5 24) or BN (n 5 61) took part in a trial

of family therapy versus guided self-care.

At baseline, eating disorder symptoms,

risk factors, and costs were assessed by

interview. Patients were reinterviewed at

6 and 12 months.

Results: Compared with EDNOS, BN

patients binged, vomited and purged sig-

nificantly more, and were more preoccu-

pied with food. Those with EDNOS had

more depression and had more current

and childhood obsessive-compulsive dis-

order. 66.6% of EDNOS versus 27.8% of

BN patients were abstinent from binge-

ing and vomiting at 1 year. Diagnosis did

not moderate treatment outcome. Costs

did not differ between groups.

Conclusion: EDNOS in adolescents is

not trival. It has milder eating disorder

symptoms but more comorbidity than

BN. VVC 2008 by Wiley Periodicals, Inc.

Keywords: bulimia nervosa; ado-

lescence; family therapy; EDNOS; guided

selfcare

(Int J Eat Disord 2008; 41:498–504)

Introduction

Three to five percent of young women sufferfrom full or partial-syndrome bulimia nervosa

(BN).1 Typically, the disorder develops in mid-to

late adolescence and adolescents now form the

majority of those who present for treatment in

primary care.2

There has recently been considerable interest inthose eating disorders that do not fulfill precisediagnostic criteria, i.e., those currently classified aseating disorders not otherwise specified (EDNOS3).These cases are the most common groups identi-fied in community samples4–6 and in out-patientsettings.7 Studies delineating different sub-types(such as purging disorder8,9), risk factors, comor-bidity, course, and outcome are only just beginningto emerge.8–11

It has been said that many of these cases are clin-ically as severely unwell as those with full syndro-

mal eating disorders,12 and an example presentedin support of this statement concerns patients whoare of normal weight, binge infrequently, but purge

frequently, which is medically dangerous.

However, an alternative possibility is that manypatients with EDNOS may have milder eatingdisorders, but are referred because of greater levels

Supported by Health Foundation, UK.

*Correspondence to: Ulrike Schmidt, MD, PhD, Section of Eating

Disorders, Institute of Psychiatry, King’s College, London, UK.

E-mail: [email protected]

Accepted 20 February 2008

1 Section of Eating Disorders, Institute of Psychiatry, King’s

College, London, United Kingdom2 Clinical Trials Unit, Institute of Psychiatry, King’s College,

London, United Kingdom3 Center for the Economics of Mental Health, Institute of

Psychiatry, King’s College, London, United Kingdom4Department of Child Psychiatry, Royal Free Hospital, London,

United Kingdom5 The Phoenix Wing Eating Disorders Service, St Ann’s Hospital,

London, United Kingdom6 Russell Unit Eating Disorders Service, Royal Free Hospital,

London, United Kingdom7 Child and Adolescent Eating Disorders Services, Frimley

Children’s Center, Frimley, United Kingdom

Published online 23 April 2008 in Wiley InterScience

(www.interscience.wiley.com). DOI: 10.1002/eat.20533

VVC 2008 Wiley Periodicals, Inc.

498 International Journal of Eating Disorders 41:6 498–504 2008

REGULAR ARTICLE

Page 2: Do adolescents with eating disorder not otherwise specified or full-syndrome bulimia nervosa differ in clinical severity, comorbidity, risk factors, treatment outcome or cost?

of comorbidity and distress. In this case, the eatingdisorder may simply be a convenient ‘‘peg’’ aroundwhich primary care physicians can organize a refer-ral. From the patient’s point of view the eating dis-order may be the most acceptable aspect of theirdifficulties. For example, someone who has a bor-derline personality disorder with major interperso-nal difficulties, who self-harms and abuses drugsand alcohol and also has weight/shape concernswith occasional binges and intermittent use of afew laxative tablets may prefer to be referred to aneating disorder service rather than a self-harm orpersonality disorder service.

In adolescents, full syndromal BN has beenfound to be less common than among youngadults13 and eating disorder symptoms may be lesssevere in the younger age group. In support of this,one study found adolescents with BN to binge lessand use laxatives less than their adult counter-parts.14 Thus, in adolescents who are developing aneating disorder, EDNOS simply may be an earlystage of the illness, before they go on to full diag-nostic criteria. There is indeed some evidence froma prospective longitudinal study in adolescentswhich shows that EDNOS cases are associated witha significant risk of developing full syndrome eatingdisorders over time.15

Adolescent patients presenting to services during

the EDNOS ‘‘stage’’ could be expected to havemilder eating disorder symptoms of shorter dura-

tion and may be those who have better family sup-port with more observant parents suggesting that

they seek help. Early treatment of these partial

cases might prevent progression to full syndromeeating disorders.

We recently conducted a randomized controlledtreatment trial (RCT) of adolescents with BN orEDNOS, comparing two different forms of psycho-logical therapy,16 namely family therapy and cogni-tive-behavioral guided self-care. Overall, guidedself-care had a slight advantage over family therapyin terms of some symptomatic outcomes, cost andacceptability.

We now want to know whether patients withEDNOS differ from those with the full syndromeof BN in terms of age or duration of their eatingdisorder, number and type of previous diagnoses,clinical severity, comorbidity, risk factors, treat-ment outcome or costs at 6 and 12 months. Wehypothesized that adolescents with EDNOSwould have a shorter duration of their eating dis-order, milder eating disorder symptoms, lowerlevels of comorbidity, lower levels of risk factorsand greater levels of family involvement com-

pared with those with full syndrome BN. We alsohypothesized that they would have a better treat-ment outcome and lower costs than those withfull-syndrome BN.

Method

Participants

Participants were all patients involved in the RCTmen-

tioned above. Participants were aged between 13 and 20

years, had been referred for treatment for BN/EDNOS to

one of the participating centers and had to have one

close other who was able to take part in their treatment.

EDNOS was defined as binge eating and/or purging

(vomiting, abuse of laxatives or diuretics) less than twice

a week or for less than 3 months, or use of inappropriate

compensatory behaviors without bingeing in patients

with normal body weight. Binge eating disorder was not

included.

Patients were randomized to either 15 sessions of fam-

ily therapy or 15 sessions of cognitive-behavioral guided

self-care treatment. In brief, the family therapy model is

an adaptation of the treatment approach that was devel-

oped at the Maudsley Hospital as a treatment for adoles-

cent anorexia nervosa.17 The emphasis is on mobilizing

family resources in helping the adolescent overcome

their eating disorder. In the early stages treatment is

problem oriented and the parents are encouraged to

take an active role in helping the young person to rees-

tablish normal patterns of eating and reduce bulimic

behaviors. Later on the impact of the eating disorder on

the family is explored and the way in which individual

psychological issues such as self esteem, individuation,

and family relationship issues may have become

entangled with the eating disorder behaviors. The final

stage of treatment focuses on ending and relapse pre-

vention.

The guided self-care treatment is based on a manual18

previously tested in adults with BN.19 There are accom-

panying workbooks for the patient and ‘‘close others’’

and a clinician’s guide.20 The therapist’s role is to guide

the patient through the manual and workbook at a pace

to fit their needs, and to answer questions and offer

advice if needed. In the early phase of treatment there is

a focus on the function of bulimia in the person’s life. In-

formation about how bulimic symptoms are maintained

is gradually introduced, using self-monitoring of

thoughts, feelings and behaviors. Problem-solving, with

behavioral experiments and goal setting are used to help

patients alter vicious cycles of behavior. A case formula-

tion is developed collaboratively. Remaining problems or

difficulties are discussed with the aim of identifying how

to work with the manual to address them. Further ses-

DO ADOLESCENTS WITH EDNOS OR BULIMIA DIFFER?

International Journal of Eating Disorders 41:6 498–504 2008 499

Page 3: Do adolescents with eating disorder not otherwise specified or full-syndrome bulimia nervosa differ in clinical severity, comorbidity, risk factors, treatment outcome or cost?

sions are devoted to relapse prevention. Regular home-

work accompanies the treatment.

Exclusion Criteria

We excluded patients with (a) a body mass index below

the 10th centile for age and sex,21 (b) those who did not

have sufficient knowledge of English to enable them to

be assessed adequately and to understand the treatment

manual and (c) those with learning disability, severe

mental illness, or alcohol/substance dependence.

We did not exclude patients on antidepressants, as in

the UK primary care physicians routinely prescribe anti-

depressants as a first step in treatment, while people wait

for specialist care.

Informed written consent was sought from subjects

and a close other (typically a parent) at initial assess-

ment. In cases aged 16 or under, consent from a par-

ent was always sought. The study was approved by the

research ethics committees of the four participating

centers.

Procedures

Trained researchers assessed participants at initial

referral, at 6 months (post-treatment) and at 12 months

using the following instruments:

1. EATATE interview (Brecelj et al., submitted): This is

a semi-structured longitudinal weight and eating

disorder history based on the longitudinal interval

follow-up examination (LIFE22) and including eat-

ing disorder examination variables.23 The EATATE

was developed in the context of a large European

study of eating disorders to assess eating disorder

symptoms longitudinally. This interview also

assesses personal and familial psychopathology

occurring prior to the index age (IA: age at which

any of the following behavioral criteria were first

met: (1) regular (at least once a month) bingeing

and/or self-induced vomiting and/or misuse of lax-

atives in order to control shape or weight; (2)

weight loss of more than 5 kg in the context of

weight or shape control; (3) in post-menarchal

females, the beginning of the absence of three con-

secutive menstrual periods in the context of weight

or shape control, across the individual’s lifetime

and at the time of assessment. A full validation of

this interview has not yet been published, but pre-

liminary analyses indicate excellent inter-rater reli-

ability with kappa values between 0.88 and 1.0 for

first and second eating disorder diagnosis and 0.82

for the number of life-time diagnoses. Spearman’s

coefficients for longitudinal assessment of behav-

ioral eating disorder symptoms are also high, e.g.,

bingeing (0.84), vomiting (0.97), laxative/diuretic

abuse (0.89), and strict dieting (0.85). We used data

from this interview for three purposes: (a) at base-

line to make DSM-diagnoses, (b) to assess eating

disorder symptoms over the previous month at

baseline, 6 months and 12 months and (c) to assess

the time course of recovery using multiple data-

points, e.g., baseline, 2, 4, 6, 8, and 10 months in

the two groups. The EATATE interview was

repeated at 6 months (i.e., after treatment) and 12

months and all patients regardless of their treat-

ment status were followed up.

2. An adapted version of the Oxford Risk Factor In-

ventory24,25: This is a retrospective interview focus-

ing on the period preceding the onset of the eating

disorder, identifying exposure to a wide range of

parental, and personal psychological and social

factors that may have placed an individual at risk

for developing an eating disorder.

3. The Client Service Receipt Inventory: This well-

established method of data collection documents

each young person’s use of education, health and

social care services, as well as additional expenses

for them or their family that are a consequence of

BN.26,27

Parental occupation, classified according to the Hol-

lingshead two-factor index of social position,28 was used

to define socio-economic status.

Self-Report Measures

A number of self-report measures on eating disorder

pathology, depression and anxiety, interpersonal and

family relationships and quality of life29–33 were also

given. With the exception of depression and anxiety these

are not the focus of this study.

Statistical Analyses

Statistical analysis was carried out using SPSS version

13. Odds ratios with 95% confidence intervals were used

for risk factor data. Statistical significance was tested

where appropriate using chi-square analyses or Fisher’s

exact test. Analyses of continuous data utilized t-tests or

Mann Whitney tests depending upon whether the

assumption of normality of distribution was met. As this

is an exploratory study no adjustments were made for

multiple testing.

Analysis of treatment outcomes was on an intention to

treat basis. To assess the interaction between diagnosis

(BN, EDNOS) and treatment group, models for repeated

measurements were fitted34 using frequencies of binge-

ing and vomiting episodes at baseline, 2, 4, 6, 8, 10, and

12 months.

Cost comparisons between the two groups were made

for three time periods (covering the 3 months prior to

SCHMIDT ET AL.

500 International Journal of Eating Disorders 41:6 498–504 2008

Page 4: Do adolescents with eating disorder not otherwise specified or full-syndrome bulimia nervosa differ in clinical severity, comorbidity, risk factors, treatment outcome or cost?

baseline, 3–6 months and 9–12 months) using independ-

ent samples t-tests with two-sided significance levels.

Results

Baseline Characteristics

The participants were 83 females (97.6%) andtwo males (2.4%) and formed two groups: 24 withEDNOS and 61 who had full-syndrome BN. Base-line characteristics are shown in Table 1. The twogroups did not differ in terms of age or duration ofeating disorder or onset of menarche. They also didnot differ in terms of proportions with a history ofanorexia nervosa or obesity. Both groups had a sim-ilar number of diagnostic changes prior to theirpresent diagnosis. The frequency of bingeing andof self-induced vomiting was significantly lower inthe EDNOS group than in the BN group, and theproportion of patients with EDNOS taking laxativesalso was lower. Patients with BN and EDNOS didnot differ on inappropriate weight and shape con-cerns or on food related fear and disgust. Patientswith BN were, however, significantly more preoccu-pied with thoughts of food. Antidepressant use wastwice as common in patients with EDNOS com-pared with patients with BN.

Comorbidity

This was assessed using the EATATE interview andincluded an assessment of premorbid and current

depression, anxiety disorders (specific phobias,social phobia), obsessive compulsive disorder, alco-hol or substance abuse/dependence. The only dif-ferences between BN and EDNOS occurred oncurrent depression, current OCD and OCD beforeindex age. [Current depression: BN: 21/58 (36.2%)vs. EDNOS: 14/23 (60.9%); odds ratio: 2.7 (95% CI1.01–7.41); current OCD: BN: 2/53 (3.8%) vs.EDNOS: 4/20 (20%); odds ratio: 6.4 (95% CI 1.1–38.1%); OCD prior to index age: BN: 1/53 (1.9%) vs.EDNOS 4/20 (20%); odds ratio: 13.0 (95% CI 1.4–125)].

Risk Factors

Risk factors were assessed using the adaptedOxford Risk factor interview. No differencesbetween groups were found. Details can beobtained from the authors on request.

Parental Involvement in Treatment

A similar proportion of patients in both groupsinvolved their parents in treatment [BN: 82.2% (44/61) versus EDNOS: 75% (18/24)].

Treatment Up-Take, Drop-Out and Outcomes

Treatment uptake was comparable in the twogroups. Nine of 61 patients with BN (14.8%) did nottake up treatment compared with 4 of 24 patientswith EDNOS (16.6%). The median number of treat-ment sessions attended in both groups was 7(range 0–17). Drop-out was defined as attendanceat less than four sessions. Attendance at four or

TABLE 1. Baseline demographic and clinical data

BN (n5 61) EDNOS (n 5 24)

Age 17.7 (SD 1.7) (range: 14–20) 17.4 (SD 1.7) (range: 14–20)Ethnicity (white vs. other) 50/60 ( 83%) 21/24 ( 87.5%)Duration 2.5 (SD 1.7) (range: 0.5–7) 2.7 (SD 2.4) (range: 0.5–9)Body mass index (kg/m2) 21.2 (SD 2.8) (range: 17.4–33.5) 20.8 (SD 1.8) (range: 18.2–25.8)Age at menarche 12.6 (SD 1.5) (range: 10–16) 12.4 (SD 1.6) (9–16)History of anorexia nervosa 10/59 (16%) 5/24 (20.8%)History of obesity 9/61 (14.8%) 1/24 (4.2%)Number of eating disorder diagnosesa 2.1 (SD 1.2) (range: 1–7) 2.5 (SD 1.4) (range: 1–6)Mean number of binges per week over previous monthb 7.2 (SD 7.0) (range: 0–24.5) 1.3 (SD 1.9) (range: 0–7)Mean number of episodes of vomiting per week over previous monthc 11.5 (SD 16.9) (0–112) 5.0 (SD 6.3) (0–21)Mean number of days of strict dieting per week over previous month 4.4 (SD 3.3) (range: 0–7) 4.1 (SD 3.5) (range: 0–7)Proportion of patients using laxativesd 12/60 (19.7%) 1/23 (4.2%)Inappropriate weight and shape concerns (max. score5 5) 4.0 (SD 1.3) (n 5 55) (range: 3–5) 4.3 (SD 0.9) (n5 23) (range: 4–5)Food-related fear & disgust (max score5 5) 3.6 (SD 1.3) (range: 1–5) 3.5 (SD 1.4) (range: 1–5)Food-related preoccupation (max.score5 3)e 1.8 (SD 0.9) (range: 0–3) 1.4 (SD 0.8) (range: 0–3)Proportion of patients on antidepressantsf 15/61 (24.5%) 14/24 (58%)HADS-anxiety score 8.9 (SD 2.7) (range: 4–15) 9.4 (SD 3.0) (range: 5–17.5)HADS-depression score 9.9 (SD 2.5) (range: 5–15) 10.5 (SD 2.9) (range: 7–18)

a Includes change from ED to no ED.b t 5 6.0, df 77.8, p\ 0.001.c t 5 2.56, df 82.9, p 5 0.012.d Chi-square 3.08, df 1, p5 0.079.e t 5 2.06, df 48.04, p5 0.04.f Chi-square 8.725, df 1, p 5 0.003.

DO ADOLESCENTS WITH EDNOS OR BULIMIA DIFFER?

International Journal of Eating Disorders 41:6 498–504 2008 501

Page 5: Do adolescents with eating disorder not otherwise specified or full-syndrome bulimia nervosa differ in clinical severity, comorbidity, risk factors, treatment outcome or cost?

more sessions of therapy was seen as a minimumadequate ‘‘dose’’ because the CBT literature inadults with BN suggests that most change occurswithin the first four sessions. The proportion ofpatients dropping out from treatment was similarfor patients with BN and EDNOS (BN: 9/61 (14.8%)and EDNOS: 3/24 (12.5%)).

Figures 1a and 1b shows the time course of re-covery for bingeing and vomiting separately foreach of the four groups. In repeated measurementanalyses, for bingeing there was a main effect oftime (F 5 8.27; df 5 6; p\ 0.0001), but no effect fordiagnostic group (BN or EDNOS) (F 5 2.6, df 5 1, p5 0.11) or treatment group (family therapy or GSC)(F 5 1.6; df 5 1; p 5 0.21). There were no interac-tions between time x diagnostic group (F 5 1.79; df5 6; p 5 0.099) or time x treatment group (F 51.38; df 5 6; p 5 0.22).

For vomiting there was also a main effect of time(F 5 7.55; df 5 6; p \ 0.0001), and no effect fordiagnostic group (BN or EDNOS) (F 5 2.17; df 5 1;p 5 0.14) or treatment group (family therapy versusguided self-care) (F 5 1.71; df 5 1; p 5 0.20). Therewere no interactions between time 3 diagnosticgroup (F 5 1.88; df 5 6; p 5 0.08) or time 3 treat-ment group (F 5 1.20; df 5 6; p5 0.30).

In addition, we also present abstinence rates(bingeing and vomiting combined) by diagnosisand treatment group (see Table 2). Although atbaseline both the EDNOS and BN group had com-parable levels of abstinence from bingeing andvomiting, at 1 year a significantly higher proportionof patients with EDNOS compared with patientswith BN achieved abstinence (12/18 vs. 10/36; Chi-square5 5.99; p5 0.01).

Cost Comparisons

There were similar levels of resource consump-tion at baseline, 6 months and 12 months for BNand EDNOS, both in terms of total public sectorcosts and expenses for the family and the youngperson. Further details on the costs are given in ourprevious study.16

Conclusion

Key findings of the present study can be summar-ized as follows: patients with EDNOS and BN didnot differ on demographic characteristics or dura-tion of eating disorder. Thus our first hypothesisthat patients with EDNOS would have a shorter du-ration of illness was not confirmed.

While by definition levels of binge eating wereexpected to differentiate between the two groups,the same is not necessarily true for levels of com-pensatory behaviors. Indeed, adolescents with BNvomited and purged significantly more than thosewith EDNOS confirming our second hypothesis.Interestingly, the two groups did not differ on the

FIGURE 1. (a) Mean weekly binge frequency in all fourgroups over the previous month. One-sided (positive ornegative) error bars are shown for greater clarity. GSC-BN,guided self care in bulimia nervosa patients; GSC-EDNOS,guided self care in patients with EDNOS; FT-BN, familytherapy in bulimia nervosa patients; FT-EDNOS, familytherapy in patients with EDNOS. (b) Mean weekly fre-quency of vomiting in all four groups over the previousmonth. At months 10 and 12, two outliers were removedin GSC-BN and one outlier was removed in EDNOS-FT.One-sided (positive or negative) error bars are shown forgreater clarity. GSC-BN, guided self care in bulimia nerv-osa patients; GSC-EDNOS, guided self care in patients withEDNOS; FT-BN, family therapy in bulimia nervosapatients; FT-EDNOS, family therapy in patients withEDNOS.

SCHMIDT ET AL.

502 International Journal of Eating Disorders 41:6 498–504 2008

Page 6: Do adolescents with eating disorder not otherwise specified or full-syndrome bulimia nervosa differ in clinical severity, comorbidity, risk factors, treatment outcome or cost?

attitudinal eating disorder symptoms (weight andshape concern) and food-related fear and disgust.

Patients with EDNOS had more current depres-sion and more current and childhood obsessive-compulsive disorder. This is the contrary of whatwe hypothesized, suggesting that patients withEDNOS may seek treatment because of theircomorbidity rather than exclusively because oftheir eating disorder symptoms.

We failed to find any difference between thegroups on childhood risk factors other than theproportion of patients with childhood-obsessivecompulsive disorder, which was higher in theEDNOS than the BN group. While this goes againstour hypothesis, which was that patients withEDNOS would have lower levels of risk factors thanthose with BN, it is in keeping with our findingsthat patients with EDNOS have more comorbiditycompared with BN.

In terms of treatment outcome, we had hypothe-sized that patients with EDNOS would have a bettertreatment outcome than those with BN. Indeed, asignificantly higher proportion of EDNOS com-pared with patients with BN were abstinent frombingeing and vomiting at follow-up. However, therewere no clear differences between EDNOS and BNon other treatment outcomes. Eating disorder diag-nosis (EDNOS or BN) did not appear to be a mod-erator of treatment effects. Having said that, thenumbers of patients in each group were very smalland the study was underpowered to detect anypotential differences in response to treatment inthe two diagnostic subgroups. Importantly, nearlytwo-thirds of patients with EDNOS were on ananti-depressant in addition to receiving psycholog-ical therapy compared with only a quarter ofpatients with BN. No differences were foundbetween groups in terms of cost to the individual,their family or the public sector. Taken together thefindings on outcome and costs suggest that EDNOSis not a trivial condition.

The strengths of the present study include theuse of a broad range of interview-based assessmentmeasures and the inclusion of a longitudinal com-ponent in the study. Weaknesses include the small

sample size and the fact that study participantswere participants in a randomized controlled trialand therefore may not be representative of the BNor EDNOS population at large.

In summary, previous research in adults hasemphasized the similarities between patients withEDNOS and BN.35 Our data suggest that there areboth differences and similarities between thesediagnostic sub-groups in adolescents. EDNOS ischaracterized by less severe behavioral eatingdisorder symptoms, similar levels of attitudinalsymptoms as BN and higher levels of comorbidity.Treatment outcomes are somewhat different inEDNOS and BN, but costs are not.

Implications for clinical practice arise from thehigher comorbidity levels in EDNOS. The severityof the comorbidity and its relationship to the eatingdisorder symptoms needs to be carefully assessed,and comorbid symptoms may need treatment intheir own right. In our study, referring general prac-titioners initiated antidepressant treatment innearly 60% of EDNOS cases. In addition, comorbidproblems may need to be included in thepsychological case formulation both as potentialvulnerability or maintaining factors for the eatingdisorder.

Finally, implications for research are that treat-ment studies including both patients with EDNOSand BN should report outcomes for both sepa-rately.

We thank all the therapists who conducted the study.

References

1. Hoek HW, van Hoeken D. Review of the prevalence and inci-

dence of eating disorders. Int J Eat Disord 2003;34:383–396.

2. Currin L, Schmidt U, Jick H, Treasure J. Time trends in the inci-

dence of bulimia nervosa. Br J Psychiatry 2005;186:132–135.

3. Norring C, Palmer. B. EDNOS, Eating Disorders Not Otherwise

Specified. Hove, East Sussex: Routledge, 2005.

4. Kjelsas E, Bjørnstrøm C, Gotestam KG. Prevalence of eating dis-

orders in female and male adolescents (14–15 years). Eat Behav

2003;5:13–25.

5. Lahortiga-Ramos F, De Irala-Estevez J, Cano-Prous A, Gual-Gar-

cıa P, Martınez-Gonzalez MA, Cervera-Enguiz S. Incidence of

TABLE 2. Abstinence rates from bingeing and vomiting by eating disorder diagnosis and treatment group

BN

Total BN

EDNOS

Total EDNOSFamily GSC Family GSC

Baseline Abstinent 1/31 (3.3%) 3/30 (10%) 4/61 (6.5%) 1/10 (10%) 1/14 (7%) 2/24 (8%)6-months Abstinent 1/25 (4%) 4/21 (19%) 5/46 (11%) 3/9 (33%) 2/10 (20%) 5/19 (26%)12-months Abstinent 6/20 (30%) 4/16 (25%) 10/36 (27.8%) 6/9 (66.6%) 6/9 (66.6%) 12/18 (66.6%)

Abstinence rates were assessed by longitudinal interview over the previous month.

DO ADOLESCENTS WITH EDNOS OR BULIMIA DIFFER?

International Journal of Eating Disorders 41:6 498–504 2008 503

Page 7: Do adolescents with eating disorder not otherwise specified or full-syndrome bulimia nervosa differ in clinical severity, comorbidity, risk factors, treatment outcome or cost?

eating disorders in Navarra (Spain). Eur Psychiatry 2005;20:

179–185.

6. Machado PP, Machado BC, Goncalves S, Hoek HW. The preva-

lence of eating disorders not otherwise specified. Int J Eat Dis-

ord 2007;40:212–217.

7. Fairburn CG, Bohn K. Eating disorder NOS (EDNOS): An example

of the troublesome ‘‘Not otherwise Specified’’ (NOS) category in

DSM-IV. Behav Res Therapy 2005;43:691–701.

8. Keel PK. Purging disorder: Subthreshold variant or full-thresh-

old eating disorder? Int J Eat Disord 2007;40 (Suppl): S89–S94.

9. Binford RB, le Grange D. Adolescents with bulimia nervosa and

eating disorder not otherwise specified-purging only. Int J Eat

Disord 2005;38:157–161.

10. Mitchell JE, Crosby RD, Wonderlich SA, Hill L, le Grange D,

Powers P, Eddy K. Latent profile analysis of a cohort of patients

with eating disorders not otherwise specified. Int J Eat Disord

2007;40 (Suppl):S95–S98.

11. Rockert W, Kaplan AS, Olmsted MP. Eating disorder not other-

wise specified: The view from a tertiary care treatment center.

Int J Eat Disord 2007;40 (Suppl):S99–S103.

12. Fairburn CG, Harrison PJ. Eating disorders. Lancet 2003;361:

407–416.

13. Nicholls D, Chater R, Lask B. Children into DSM don’t go: A

comparison of classification systems for eating disorders in

childhood and early adolescence. Int J Eat Disord 2000;28:317–

324.

14. Fisher M, Schneider M, Burns J, Symons H, Mandel F. Differences

between adolescents and young adults at presentation to an

eating disorders program. J Adolesc Health 2001;28:222–227.

15. Striegel-Moore RH, Seeley JR, Lewinsohn PM. Psychosocial

adjustment in young adulthood of women who experienced an

eating disorder during adolescence. Am Acad Child Adolesc Psy-

chiatry 2003;42:587–593.

16. Schmidt U, Lee S, Beecham J, Perkins S, Treasure J, Yi I, et al. A

randomized controlled trial of family therapy and cognitive

behavior therapy guided self-care for adolescents with bulimia

nervosa and related disorders. Am J Psychiatry 2007;164:591–

598.

17. Eisler I. The empirical and theoretical base of family therapy

and multiple family day therapy for adolescent anorexia nerv-

osa. J Fam Ther 2005;27:104–131.

18. Schmidt U, Treasure J. Getting better bit(e) by bit(e). A treat-

ment manual for sufferers of bulimia nervosa. Hove, East Sus-

sex: Psychology Press, 1997.

19. Perkins SJ, Murphy R, Schmidt U, Williams C. Self-help and

guided self-help for eating disorders. Cochrane Database of Sys-

tematic Reviews, 3, CD004191, 2006.

20. Treasure J, Schmidt U. The Clinicians Guide to Getting Better

Bit(e) by Bit(e). Hove, East Sussex: Psychology Press, 1997.

21. Cole TJ, Freeman JV, Preece MA. Body mass index reference

curves for the UK, 1990. Arch Dis Child 1995;73:25–29.

22. Keller MB, Lavari PW, Friedman B, Nielsen E, Endicott J, McDo-

nald-Scott P, et al. The longitudinal interval follow-up evalua-

tion. A comprehensive method for assessment of outcome in

prospective longitudinal studies. Arch Gen Psychiatry 1987;44:

540–548.

23. Cooper Z, Fairburn CG. The eating disorders examination. In:

Fairburn CG, Wilson GT, editors. Binge Eating: Nature, Assess-

ment and Treatment, 12th ed. New York: Guilford Press, 1993,

pp 317–360.

24. Fairburn CG, Welch SL, Doll HA, Davies BA, O’Connor ME. Risk

factors for bulimia nervosa. A community-based case-control

study. Arch Gen Psychiatry 1997;54:509–517.

25. Fairburn CG, Doll HA, Welch SL, Hay PJ, Davies BA, O’Connor

ME. Risk factors for binge eating disorder: A community-based

case-control study. Arch Gen Psychiatry 1998;55:425–432.

26. Beecham J, Knapp M. Costing psychiatric interventions. In:

Thornicroft G, Brewin C, Wing JK, editors. Measuring Mental

Health Needs. London, UK: Gaskell, 1992, pp. 163–183.

27. Beecham J. Collecting and estimating costs. In: Knapp M, edi-

tor. The Economic Evaluation of Mental Health. Aldershot, UK:

Arena, 1995, pp. 157–174.

28. Guy W. ECDEU Assessment Manual for Psychopharmocology.

Rockville, MD: US Department of Health, Education and Wel-

fare, 516–520.

29. Bauer S, Winn S, Schmidt S, Kordy H. Construction, scoring and

validation of the short evaluation of eating disorders (SEED).

Eur Eat Disord Rev 1995;13:191–200.

30. Zigmond AS, Snaith RP. The hospital anxiety and depression

scale. Acta Psychiatr Scand 1983;67:361–370.

31. Horowitz LM, Rosenberg AE, Baer BA, Ureno G, Villasenor V. In-

ventory of interpersonal problems: Psychometric properties

and clinical applications. J Consult Clin Psychol 1988;56:885–

892.

32. Cole JD, Kazarian SS. The level of expressed emotion scale: A

new measure of expressed emotion. J Clin Psychol 1988;44:

392–397.

33. Brooks R, with the EuroQol Group. EuroQol: The current state

of play. Health Policy 1996;37:53–72.

34. Lindsay JK. Models for Repeated Measurements. Oxford Statisti-

cal Science Series 10. Oxford, UK: Clarendon Press.

35. Le Grange D, Binford RB, Peterson CB, Crow SJ, Crosby RD, Klein

MH, et al. DSM-IV threshold versus subthreshold bulimia nerv-

osa. Int J Eat Disord 2006;39:462–467.

SCHMIDT ET AL.

504 International Journal of Eating Disorders 41:6 498–504 2008