Severe anorexia nervosa in men: Comparison with severe AN in women and analysis of mortality

Severe Anorexia Nervosa in Men: Comparison withSevere AN in Women and Analysis of Mortality

Juliette Gueguen, MD1,2

Nathalie Godart, MD, PhD1,2,3

Jean Chambry, MD1,2,4

Annick Brun-Eberentz, MD5

Christine Foulon, MD1,2,5

Snezana M. Divac, PhD5

Julien-Daniel Guelfi, MD5

Frederic Rouillon, MD2,5,6

Bruno Falissard, MD, PhD1,2

Caroline Huas, MD, PhD1,2,7*

ABSTRACT

Objective: To compare clinical charac-

teristics of men and women with severe

AN and to analyze mortality in men.

Method: One thousand and nine

patients including 23 anorectic males

were hospitalized in St. Anne Hospital in

Paris between 1988 and 2004. Data were

collected during hospitalization. Fatal

outcome was assessed in 2008.

Results: Men presented significantly

later age of onset, were more likely to

have a history of premorbid overweight

than women and less likely to have

attempted suicide. Mortality in men was

high (standardized mortality ratio: 8.08;

95% CI: 1.62–23.62). Several predictive

factors for mortality in men were identi-

fied: lower admission body mass index

(BMI), later age at admission, and AN-R

subtype. All the three deceased patients

had dropped out from the inpatient unit.

The 10-year survival did not differ

between men and women, but men died

sooner after hospitalization.

Discussion: Male inpatients should

receive close follow-up after their discharge,

especially if they have a restrictive form of

AN, present low BMI, or are older at admis-

sion.VVC 2012 by Wiley Periodicals, Inc.

Keywords: eating disorders/epidemiology;

eating disorders/mortality; follow-up;

studies; prognosis; male; hospitalization;

survival analysis; sex; factors; adult;

anorexia nervosa

(Int J Eat Disord 2012; 45:537–545)

Introduction

Eating disorders occur mostly in females, with afemale–male ratio of 10:1 for anorexia nervosa(AN).1,2 Males with AN present many similaritieswith their female counterparts. However, a laterage of onset3,4 and a history of premorbid over-weight5–7 have been reported for males. Compari-sons in terms of current eating features (using theEating Disorders Inventory) have produced dis-cordant findings.3,8,9

There are few studies focusing on AN in malesbecause of the very small sample sizes. The largest

series describing or comparing male patients withED included 135 patients.5 For subjects recruited in

tertiary centers only, the largest series included 79

patients.10 These studies were not prospective and

did not investigate prognostic factors or mortality. To

our knowledge, only one study has previously calcu-

lated an 8.2 (95% CI: 2.7–19.1) standardized mortality

ratio (SMR) in hospitalized males (n 5 63) with a

main or secondary diagnosis of anorexia.11 This study

was retrospective and based on nation-wide register

with very little clinical data. No predictive factors for

mortality were described. No SMR has ever been cal-

culated in severe patients hospitalized in tertiary cen-

ters. However, inpatients in tertiary centers are the

most ill and at very great risk of death, which is why

this is an important population to study.

In addition, most of the follow-up studies inmale patients reported a maximum 1-year follow-up. To our knowledge, only one male populationwas followed up for up to 8 years on average. Itenabled description of prognostic factors for pooroutcome in males12 and comparison with prognos-tic factors in females.13 Later age at presentation,longer duration of illness, previous treatment,greater weight loss, absence of premorbid sexualactivity, poor relationship with parents duringchildhood, lack of social involvement, and difficultyexpressing emotions were reported to negatively

Accepted 16 October 2011

1 Inserm, U669, Paris, France2 Univ Paris-Sud and Univ Paris Descartes, UMR-S0669, Paris,

France3 Institut Mutualiste Montsouris, Paris, France4 AP-HP, Hopital Kremlin Bicetre, Department of

Pedospychiatry, Kremlin Bicetre, France5 Hopital Sainte-Anne, Clinique des maladies mentales et de

l’encephale, Paris, France6 Center Psychiatry and Neurosciences Inserm U894, Paris,

France7 Univ Tours, Department of General Practice, 37000 Tours,

France

*Correspondence to: Dr. Caroline Huas, Univ Paris-Sud and Univ

Paris Descartes, UMR-S0669, Paris. E-mail: [email protected]

Published online 24 January 2012 in Wiley Online Library

(wileyonlinelibrary.com). DOI: 10.1002/eat.20987

VVC 2012 Wiley Periodicals, Inc.

International Journal of Eating Disorders 45:4 537–545 2012 537

REGULAR ARTICLE

affect prognosis in males with ED,10,12 and it can bethought that they would also affect mortality. Incontrast, impulsivity-related behaviors (such asvomiting/binging) were found to be associated withgood outcome in males, in contrast to females.13

The purposes of this study were as follows:

� To describe a sample of 23 male patients hos-pitalized for AN in terms of sociodemographic,clinical, psychological, and outcome data andto compare with females with AN.

� To analyze mortality in male patients with AN,to provide information on potential predictivefactors for mortality in men, and to comparesurvival according to gender.

Method and Material

Participants

The original sample comprised 1009 patients (41males and 968 females) with DSM-IV criteria for ED,either restrictive AN (AN-R), binging/purging AN(AN-B/P), bulimia nervosa, or eating disorder nototherwise specified. The patients were recruitedwhile hospitalized for the first time in the EatingDisorder Unit of the Clinique des Maladies Mentaleset de l’Encephale at Sainte-Anne Hospital, Paris,France, between January 1988 and July 2004. Ano-rectic patients were expected to reach a body massindex (BMI) of 20 for females and of 21 for males atthe end of hospitalization (details of hospitalizationand treatment described in Refs. 14 and 15). Overall,23 males and 601 females were hospitalized for AN.

Data collection started at the opening of the EDunit. The moment of inclusion in the study (whichrepresents the beginning of mortality follow-up)was defined as the first admission to the unit dur-ing that period. The endpoint of the mortality sta-tus research was the February 7, 2008 (date of deathdata collection).

Procedures

This work was accepted by the French NationalCommittee for private freedoms CNIL (CommissionNationale de l’Informatique et des Libertes) and byan independent review board (CCTIRS, ComiteConsultatif sur le Traitement de l’Information enmatiere de Recherche dans le domaine de la Sante).The standardized evaluation performed was part ofthe regular admission procedures. Patients were

informed that data were to be used in future stud-ies, and verbal consent was obtained. The exclusioncriteria were patient refusal and inability to under-stand and read French and to complete forms andquestionnaires. No patients met exclusion criteriaduring the recruitment period, and no patientsrefused to participate.

Assessment

At Admission: Sociodemographic, Personal, Clinical, andPsychological Data. As previously described,14 allinpatients completed questionnaires and wereassessed by trained psychiatrists or psychologistsfor the purpose of individualizing treatment.

Data collected comprised:

� DSM-IV diagnosis was established in twostages. First, at the end of the hospitalization,the two doctors of the unit met and togethercoded the pathology in both the ICD (9 then10) and DSM (III-R then IV) systems. This pro-cedure was consistent throughout the datacollection. Then, the diagnoses established forDSM III-R were recoded post hoc in DSM-IVdiagnoses with the doctor in charge of thedepartment for the full duration of data collec-tion (CF) and if needed with reference to themedical files. It can also be noted that thehead of the unit (Pr Guelfi) was coordinator ofthe French translation of DSM-IV.

� Clinical quantitative data: age at onset, age atadmission, duration of illness, duration of hos-pitalization in days, BMI at admission, mini-mum and maximum BMI since puberty, anddesired BMI (defined from the answer to thefollowing question ‘‘What do you think yourweight should be after treatment?’’).

� Psychometric data: four self-report question-naires were administered: the Symptom CheckList-90 revised (SCL-90R16), the Eating Disor-der Inventory (EDI17), the Eating Attitude Test40 (EAT-4018), and the 13-item Beck Depres-sion Inventory (BDI19). The global severityindex (GSI) and subscale scores on the SCL-90R can vary from 0 to 4; the global score onthe EDI, BDI, and EAT, and subscale scores onthe SCL-90R and EDI were also calculated.

� Sociodemographic and personal data: highereducation, being with a partner, having chil-dren, and having friends.

� Personal data regarding love and sexuallife: being involved in a romantic relationship,history of sexual relationship, and sexualsatisfaction.

GUEGUEN ET AL.

538 International Journal of Eating Disorders 45:4 537–545 2012

� Past medical history: history of premorbidoverweight (defined as a history of maximumBMI � 25 kg/m2), history of previous hospital-ization for ED, and history of suicide attempt.

� Clinical features: occurrence of rumination,water intake in large quantity, laxative use, di-uretic use, diet pill use, and regular consump-tion of tobacco, alcohol, or other drugs.

During follow-up, data collected comprised:

� At discharge: dropout, defined in Ref. 15.

� In February 2008: fatal outcome was obtainedfrom the National Institute of Statistics and Eco-nomics Studies (Institut National de la Statistiqueet des Etudes Economiques; INSEE). The INSEEidentification process was performed usingname, surname, date, and place of birth. Vitalstatus was ascertained for all patients. Causes ofdeath were obtained from Centre d’Epidemiologiesur les causes medicales de Deces (the French epi-demiological center collecting data on causes ofdeath) and from the medical staff.

Factors Predictive of Mortality. The variables testedwere either factors reported to be significantlylinked to death in females with ED in the literature(purging behaviors, vomiting, previous hospitaliza-tion, low minimum and admission BMI, later age atonset and at admission, duration of illness, and sui-cide attempt20), or factors reported to be associatedwith poor prognosis in males with ED in the litera-ture (late age at admission, low-minimum BMI, nohistory of sexual relationship, and absence of purg-ing), or factors to test new clinical hypotheses (ANsubtype, high global and subscale SCL and EDIscores, and absence of friendly relationships).

We chose to transform quantitative variables intoclass variables to make results easy to use in clini-cal practice: age at admission, duration of illness,EDI, and SCL scores were divided into two groupsaccording to the median, age at onset was dividedinto two groups according to the third quartile,admission BMI was recoded as \15 or �15 kg/m2

(BMI \ 15 indicates a moderately severe form21),and minimum BMI was recoded as\13 or�13 kg/m2

(BMI \ 13 is associated with a higher risk ofdeath21).

Statistical Analysis

Analyses were performed using R 2.8.1 soft-ware.22 Type one error for statistical tests of

hypothesis was equal to 0.05. Nonparametric testswere used when necessary (Fisher exact test andWilcoxon test). No adjustments for multiple testingwere made.23

Descriptive and Comparative Analysis. The anorecticmale patients are first described in terms of socio-demographic, personal, clinical, psychological, andoutcome data (see variables of interest), and theresults are also presented by AN subtype (AN-R 5 9and AN-B/P 5 14). Then, anorectic males are com-pared to their female counterparts (n 5 601).

Crude Mortality Rate and SMR. The CMR was calcu-lated as usual by dividing the number of deaths bythe total number in the cohort that was traced.Standardized mortality ratio (SMR) calculation wasperformed using indirect methods.24 The expectednumber of deaths was obtained by applying age,gender, and 5-year-specific mortalities for the gen-eral French population (obtained from INSEE) tothe corresponding cumulative person-year in thestudy cohort. We calculated two SMR and theirrelated confidence intervals:24 one for all AN casesand one for the AN-R subgroup.

Factors Predictive of Mortality—Bivariate Analysis. Thevariables were divided into three groups: five varia-bles were ‘‘admission’’ predictors (i.e., factors pres-ent at admission), nine were ‘‘lifetime’’ predictors(i.e., reported present at any time during life), andone was a ‘‘discharge’’ predictor (drop out; seeTable 3).

Survival curves were compared using the log-rank test.25 Given the small sample size and thesmall number of events (death, n 5 3), no multivar-iate analysis was performed.

Survival and Gender. Survival curves were comparedusing the log-rank test. Six hundred and onefemales with AN were hospitalized during the pe-riod and five hundred and thirty-nine were fol-lowed up.20

Results

Male Patient Characteristics and Comparison

with Females

AN-B/P was the most frequent diagnosis amongmen, affecting 14 patients (61.0%).

Clinical, psychometric, personal, and sociode-mographic data are described by diagnostic sub-group in Table 1. Briefly, mean age was 26.6 years(SD 5 6.1) at admission and 20.8 years (SD 5 4.3)at onset. More than half reported previous hospital-izations for ED. Premorbid overweight was frequent

MORTALITY IN MALES WITH AN


(39%). As expected, dropout patients had a signifi-cantly shorter duration of hospitalization and a sig-nificantly lower discharge BMI.

The 23 men with AN were compared to 601women with AN. The description of the 601 womenwith AN is given in Table 1 and also available in apreviously published article.20 Significant differen-ces between men and women were observed. Menpresented later age of onset (20.78 years vs. 18.06years; p\ 0.001) were more likely to have a historyof premorbid overweight (39% vs. 13%, p 5 0.002)and less likely to have a history of suicide attempt(4% vs. 29%; p 5 0.01). Their admission, minimum,maximum, and desired BMI were higher (15.58 vs.

14.47, p 5 0.007; 14.21 vs. 13.32, p 5 0.012; 24.07 vs.21.49, p\ 0.001; 19.38 kg/m2 vs. 17.61 kg/m2, p\0.001). Duration of hospitalization was significantlyshorter in males (56.6 vs. 73.7 days, p 5 0.027),even when the therapeutic program was completed(84.44 vs. 107.91 days, p 5 0.017). No other signifi-cant differences were found concerning clinical,sociodemographic, and personal data. However,the duration of illness tended to be shorter in men(5.91 vs. 8.38 years, p 5 0.08), certain clinical fea-tures tended to be more frequent in men: rumina-tion (30 vs. 16%, p 5 0.08), and their relationshipprofile tended to be different: males with ANtended to be less likely to report friendly relation-

TABLE 1. Sociodemographic and clinical characteristics in male and female patients with AN

All AN AN-R AN-B/P

Males Females Males Females Males FemalesN5 23a N5 601a N5 9a N5 320a N5 14a N5 281a

Clinical data Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD) Mean (SD)Age of onset (years) 20.8 (4.25) 18.1 (4.66) 20.8 (4.92) 18.5 (5.1) 20.8 (3.96) 17.5 (4.1)Age at admission (years) 26.59 (6.13) 26.44 (7.33) 29.4 (8.17) 25.9 (7.6) 24.79 (3.68) 27.0 (7.0)Duration of illness (years) 5.91 (5.75) 8.38 (7.41) 8.55 (7.68) 7.4 (7.4) 4.21 (3.42) 9.5 (7.3)BMI at admission (kg/m2) 15.59 (1.90) 14.47 (1.76) 14.85 (2.11) 13.9 (1.7) 16.06 (1.67) 15.1 (1.7)Minimum BMI (kg/m2) 14.21 (2.07) 13.32 (1.99) 13.15 (2.03) 12.9 (1.8) 14.88 (1.85) 13.8 (2.1)Maximum BMI (kg/m2) 24.07 (3.56) 21.49 (3.19) 23.86 (3.84) 21.2 (3.03) 24.21 (3.51) 21.8 (3.34)Desired BMI (kg/m2) 19.38 (2.78) 17.61 (1.49) 17.6 (2.9) 17.8 (1.44) 20.4 (2.22) 17.4 (1.53)Discharge BMI (kg/m2) 18.52 (2.37) 18.3 (2.21) 18.09 (2.72) 18.2 (2.29) 18.72 (2.28) 18.36 (2.11)Treatment completed 20.5 (1.22) 19.6 (0.86) 20.47 (0.26) 19.55 (0.97) 20.49 (1.53) 19.7 (0.79)

Dropout 16.8 (1.62) 16.3 (2.08) 15.71 (1.23) 15.98 (2.09) 17.20 (1.62) 16.7 (2.02)Psychometric dataSCL-90R: GSI 1.23 (0.75) 1.54 (0.67) 1.11 (0.39) 0.15 (0.06) 1.32 (0.94) 0.18 (0.07)BDI 15.4 (6.68) 17.9 (7.44) 13.1 (4.78) 16.1 (6.8) 17.0 (7.49) 19.9 (7.6)EAT 52.5 (23.89) 57.4 (20.61) 49.0 (23.16) 52.7 (21.4) 54.9 (25.01) 62.6 (18.4)EDI 76.0 (29.31) 78.5 (27.88) 63.6 (21.36) 65.6 (23.6) 83.5 (31.65) 93.0 (25.1)

Duration of hospitalization (days) 56.7 (33.75) 73.7 (45.73) 65.1 (34.2) 82.5 (48.1) 51.2 (33.6) 62.4 (40.6)Sociodemographic data % (n) % (n) % (n) % (n) % (n) % (n)Higher education 65.2 (15) 64.2 (380) 88.9 (8) 64.9 (203) 50.0 (7) 63.4 (177)Not being single 4.3 (1) 19.9 (119) 11.1 (1) 18.2 (58) 0 (0) 21.8(61)Having children 4.3 (1) 12.4 (71) 11.1 (1) 9.1(28) 0 (0) 16.2(43)Having friends 68.2 (15) 84.1 (490) 44.4 (4) 89.6 (275) 84.6 (11) 77.9 (215)

Personal dataHistory of romantic relationship 22.7 (5) 41.2 (238) 11.1 (1) 38.6 (118) 30.8 (4) 44.1 (120)History of sexual relationship 56.5 (13) 67.9 (387) 33.3 (3) 59.5 (181) 71.4 (10) 77.5 (206)Sexual satisfaction 9.1 (2) 15.7 (77) 44.4 (4) 10.6 (34) 35.7 (5) 15.3 (43)Previous hospitalization 63.6 (14) 60.0 (354) 77.8 (7) 62.4 (183) 53.8 (7) 57.9 (171)Suicide attempt 4.3 (1) 28.9 (1.70) 0 17.5 (55) 7.1 (1) 42.1 (115)

Clinical traitsBinging 61 (14) 53.2 (311) 11 (1) 17.7 (55) 93 (13) 93.4 (256)Vomiting 65 (15) 35.7 (209) 22 (2) 17.7 (55) 93 (13) 56.2 (154)Rumination 30.4 (7) 15.9 (92) 22.2 (2) 12.0 (37) 35.7 (5) 20.4 (55)Water intake in large quantity 65.2 (15) 48 (276) 44.4 (4) 37.8 (116) 78.6 (11) 59.0 (160)Laxative use 26.1 (6) 36.3 (212) 11.1 (1) 29.9 (93) 35.7 (5) 43.6 (119)Diuretic use 8.7 (2) 7.0 (42) 0 (0) 5.2 (16) 14.3 (2) 9.5 (26)Diet pill use 0 6.0 (35) 0 3.6 (11) 0 8.8 (24)Tobacco use 34.8 (8) 45.0 (260) 22.2 (2) 36.7 (113) 42.9 (6) 54.4 (147)Alcohol abuse 4.3 (1) 14.0 (79) 0 8.1 (25) 7.1 (1) 20.2 (54)Drug use 8.7 (2) 8.0 (46) 0 4.3 (13) 14.3 (2) 12.4 (33)Dropout 60.9 (14) 52.9 (318) 66.7 (6) 50.0 (160) 57.1 (8) 56.2 (158)

Characteristics are in given in bold when there is a significant difference between male and female AN patients.aN may vary, due to missing data.AN, anorexia nervosa; AN-R, anorexia nervosa restrictive type; AN-B/P, anorexia nervosa binging/purging type; BN, bulimia nervosa; ED-NOS, eating dis-orders not otherwise specified; BMI, body mass index; BDI, Beck Depression Inventory; EAT, Eating Attitude Test; EDI, Eating Disorder Inventory; SCL-90RGSI, Symptom Check list-90 revised Global Severity Index; SD, standard deviation.

GUEGUEN ET AL.


ships (68 vs. 84%, p 5 0.07), to report a romanticrelationship (23 vs. 41%, p 5 0.08), and to be livingwith a partner (4 vs. 20%, p 5 0.10) than theirfemale counterparts. The absence of significantresults on these points could be due to a lack ofpower. Concerning psychometric data, only theSCL-R GSI and four of its subscales scores (somati-zation, anxiety, interpersonal sensitivity, and addi-tional items) were significantly lower in males (seeTable 2).

Mortality in Men

Three deaths occurred among the males. They wereall AN-R patients, and no AN-B/P patient died duringfollow-up. The CMRwas 13.04% for the overall sample.

The average duration of follow-up was 9.8 yearswith a total of 225.6 person-years. The expectednumber of deaths was 0.37. The SMR for the ANmales was 8.08 (95% CI: 1.62–23.62), and, for theAN-R subgroup, it was 13.19 (95% CI: 2.65–38.55).

The median duration between inclusion and deathwas 1 year (0.43–2.01 years); all deaths occurred inthe first 2 years of follow-up. The mean age at thetime of death was 33.28 years (range, 28–40.1 years).The three recorded deaths were due to medical com-plications of anorexia. The three deceased patientsall dropped out (3 of 14 dropouts: 21.4%).

Predictors of Mortality in Men—Bivariate

Analysis

The following admission factors were found to besignificantly associated with fatal outcome: later ageat admission (�27 years), BMI at admission \ 15(kg/m2), and AN restrictive subtype (see Table 3).

Survival and Gender

Long-term survival did not differ between menand women, but 3-year survival differed signifi-cantly, males dying sooner after discharge thanfemales (p5 0.017; see Fig. 1).

TABLE 2. Summary of SCL-90R (GSI and subscales), BDI,EAT, and EDI scores in anorectic male and femalepatients (population 2)

Males(N5 23a)

FemalesN5 601a

Analysis(Wilcoxon Test)

Mean (SD) Mean (SD) p Value

SCL-GSI 1.23 (0.75) 1.54 (0.68) .019Somatization 1.01 (0.78) 1.45 (0.84) .013Obsessive compulsive 1.41 (0.88) 1.67 (0.93) .19Interpersonal sensitivity 1.39 (0.86) 1.87 (0.90) .007Depression 1.90 (0.93) 2.12 (0.89) .21Anxiety 1.13 (1.02) 1.54 (0.91) .022Hostility 0.98 (0.87) 1.10 (0.80) .36Phobic anxiety 0.66 (0.88) 0.88 (0.83) .13Paranoid ideation 1.06 (0.83) 1.24 (0.85) .31Psychoticism 0.98 (0.73) 1.09 (0.68) .31Additional items 1.54 (0.81) 1.96 (0.85) .02

BDI 15.41 (6.68) 17.88 (7.44) .09EAT 52.5 (23.89) 57.35 (20.61) .35EDI 75.95 (29.31) 78.45 (27.88) .79

a n may vary, due to missing data.BDI, Beck Depression Inventory; EAT, Eating Attitude Test; EDI, EatingDisorder Inventory; SCL-90R GSI, Symptom Check list-90 revised GlobalSeverity Index; SD, standard deviation.In bold: significant differences between men and women.

TABLE 3. Comparison of living and deceased patients (AN males)

Alive (N5 20) Deceased (N5 3) Analysis (Log Rank)Characteristics % (N/nr)a % (N/nr) p Value

Admission factorsAge at admission � 26 years 35 (7/20) 100 (3/3) .0369BMI at admission\15 (kg/m2) 25 (5/20) 100 (3/3) .0109AN restrictive subtype 30 (6/20) 100 (3/3) .0215SCL-90R subscale scores at admissionb:Interpersonal sensitivity score � median 47.06 (8/17) 100 (3/3) .0995Paranoid ideation score � median 55.55 (10/18) 100 (3/3) .156

EDI global score � medianc 55.55 (10/18) 33.33 (1/3) .525Lifetime factorsAge of onset � 24 years 20 (4/20) 66.67 (2/3) .0676Minimum BMI\ 13 (kg/m2) 15 (3/20) 66.67 (2/3) .0536Friendly relationships 73.68 (14/19) 33.33 (1/3) .142History of sexual relationship 55 (11/20) 33.33 (1/3) .395Previous hospitalizations 63.16 (12/19) 66.67 (2/3) .863Duration of illness[ 4 years 45 (9/20) 66.67 (2/3) .49Suicide attempt 5 (1/20) 0 (0/3) .705Vomiting 75 (15/20) 0 (0/3) .0109Purging behavior (vomiting, use of laxatives,or diuretics)

75 (15/20) 0 (0/3) .0109

Discharge factorsDropout (11/20) (3/3) .149

a nr, number of responses.b GSI was not significant; only significant results for subscales are reported in the table.c No subscale scores were significant.SCL-90R, check list-90 revised; EDI, Eating Disorder Inventory; BMI, Body Mass Index.



Discussion

This study provides new information on very severeforms of AN in males recruited during hospitaliza-tion in a tertiary center. It concerns their character-istics in comparison with women and their mortal-ity rates. This is the first study to investigate predic-tive factors of mortality in men, comparing menand women. This is, to our knowledge, the secondstudy to calculate a SMR in males with AN and thefirst in males with such a severe form of AN. Com-pared to females, male patients with AN were sig-nificantly older at onset and were more likely tohave a history of premorbid overweight and lesslikely to have a history of suicide attempt. Theobserved mortality risk was very high. Ten years onaverage after their hospitalization, the mortalityrate was eight times higher compared to the Frenchgeneral population of the same gender and age(SMR 5 8.08). Mortality risk was even higheramong AN-R patients (SMR 5 13.19). The threedeceased patients died on average 1 year sooner af-ter hospitalization than women. All three weredropouts.

Comparison Between Males and

Females with AN

Among males, the later age of onset (20.78 years)is consistent with previous studies,3,10 as is the fre-quency of premorbid overweight.5–7

Survival did not differ between males andfemales with anorexia nervosa (AN) when long-term survival was considered (10 years on average).

However, it did differ when 3-year survival wasconsidered, indicating that men die sooner afterhospitalization.

We found very similar eating features in malesand females, with similar clinical presentation andno differences in the EDI scores, these results beingconsistent with previous studies.3,5,6,10,26

Most of the differences observed could simplyreflect a gender effect. Thus, we found differenceson the GSI and on interpersonal sensitivity, somati-zation, anxiety, and additional items subscales inthe SCL-90R, with males presenting lower scores.In the community, Sepulveda27 also reported lowerscores for male students for the GSI, for the inter-personal sensitivity, somatization, and anxiety sub-scales, as well as for the depression subscale on theSCL-90. We also found higher BMI in males than infemales, as expected in the general population,28

and as correlates all other BMI (minimal, maxi-mum, and desired) are higher. We also found thatmales with AN were less likely to have a history ofsuicide attempt, which is in line with the data onthe general population.29

It seems that men with AN could be less involvedin friendly and romantic relationships. Male andfemale patients with AN could differ in terms ofinterpersonal interaction, social investment, andsupportive relationships. Males could be embar-rassed at suffering from a disease that occursmostly in females. Andersen30 also noted thatmany males with ED were uncomfortable with thestereotyped role society that expects men toassume but that they could not come to terms withtheir difference.30 The fear of being considered asunmasculine or ridiculous could account for theirpoor social involvement. It could also be that, com-pared to women with AN, men with AN presentmore psychiatric comorbidities that interfere withthe ability to interact with others, such as psychoticdisorders.31 These trend results need to be con-firmed by further studies.

We found later age of onset and more frequentpremorbid overweight in males, which are theresults previously reported in the literature.3–7 Thelater age of onset could be partly explained by thelater onset of puberty in males.32 The more fre-quent premorbid obesity in males could beexplained in different ways. First, the social pres-sure concerning body weight is less intense inmales than in females,33 and females with AN havebeen shown to present more weight concern thantheir male counterparts.26 Second, overweight canlead to teasing from family and peers, and teasinghas been described in males as a risk factor forengaging in restrictive eating behaviors.33,34

FIGURE 1. Kaplan–Meier estimate survival by gender.

GUEGUEN ET AL.


Dropout was not more frequent among malesthan females. However, duration of hospitalizationwas significantly shorter in males, indicating thatthey reached their weight objectives more quickly.This does not indicate a faster treatment: weightfluctuations are more frequent among males withED.35,36 Some male patients may not enjoy beingsurrounded mostly by women. It is therefore possi-ble that some male patients reach their weightobjectives merely in order to get out as quickly aspossible from an environment in which they feeluncomfortable.

Mortality

The SMR observed for the AN (8.08; 95 CI: 1.62–23.62) is very similar to that calculated by Møller–Madsen in hospitalized males with a main or sec-ondary diagnosis of AN (8.2).11 This study confirmsthat mortality associated with ED is high in hospi-talized males, as it is in females.11,37 SMR for thefemale AN population of our study is high or evenhigher (10.6; 95 CI: 7.6–14.4).20

The restrictive type of AN was found to be associ-ated with higher mortality for the first time. This isconsistent with the hypothesis that, in contrast towomen (among whom the presence of vomitinghas been reported as a poor prognosis factor38), thepresence of vomiting is a positive factor in males.13

Some factors predictive for mortality in men aresimilar to the previous findings in female patients(later age at admission and lower admissionBMI).39–42 Lower admission BMI indicates severemalnutrition and is therefore associated withgreater risk of medical complications. Later age atadmission could either be associated with later ageof onset or later access to healthcare in case of afirst hospitalization. In case of repeated hospitali-zation, it could be a sign of a longer duration ofevolution, a factor previously described as predic-tive of death in females.43

The occurrence of male deaths over a short pe-riod of time after hospitalization (\2.5 years) fol-lows the pattern described by Nielsen.37 The risk ofdeath overtime after hospitalization differedbetween males and females: males seemed to diesoon after hospitalization, whereas women diedover a longer period of time, again as described byNielsen.37 Several hypotheses can be made in orderto explain why men die in a shorter period of timeafter hospitalization than women. The threedeceased men had dropped out from inpatient unitat very low weights and died from AN complica-tions. It could be that men dropped out in an

acuter stage of the disease and had less social sup-port than women to help them to seek treatment.

Another question is whether the tools used toassess the severity of the disorders in males weresuitable. Most of the tools were developed forwomen. We may need to reconsider the assessmenttools for males. In males, the EDI is less reliable,44

and the use of tools designed for males, taking intoaccount certain specific concerns (such as muscu-larity, physical exercise, and steroid use), has beenpreviously recommended.45–47 Finally, hospitaliza-tion may provide fewer benefits in men comparedto women, and the posthospitalization period maybe very critical for them, because they have lesssocial support.

Strengths and Limitations

We only considered here severe adult patientshospitalized in a tertiary center, so that generaliza-tion of results to all AN patients is not justifiable.Moreover, the small sample size did not enablemultivariate analysis, which would be of great im-portance to confirm our results. The fact that alldeaths occurred for AN-R patient needs to be con-firmed in further studies. The study is also limitedby the absence of examination of features shown orthought to be associated with gender in eating dis-orders, namely sexual orientation and excessiveexercise. The recruitment of subjects from a singlecenter, the high-inclusion rate (no refusal), and thehigh-follow-up rate for male patients (100%) allconstitute strengths of this study. The identificationof potential predictive factors for mortality is veryuseful in clinical practice as it can enable the iden-tification of very severe forms and lead to specifictreatment and follow-up.

Conclusion

Men with AN have a later age of onset and a morefrequent history of premorbid overweight thanwomen. Mortality among male patients with severeforms of ED is high. Their posthospitalization mor-tality is high and occurs soon after discharge.

The potential predictive factors we identified formortality will have to be further analyzed throughnew studies with larger samples enabling multivar-iate models.

Patients suffering of a restrictive form of ANincurred a greater risk of death when they pre-sented: a lower BMI at admission or discharge(dropouts), older age at admission, and poor social



support. Some of these factors have been previ-ously documented in women, such as factors indi-cating severe malnutrition (BMI at admission \15) or later age at admission. Some of them couldbe more specific to men, such as factors indicatinga restrictive type (AN-R subgroup). In addition,men died sooner after hospitalization. Thisimplies that a very careful follow-up should beimplemented for these very severe patients, espe-cially for those who drop out and those presentinga restrictive or nonpurging form or a low-socialsupport. Follow-up would be useful to screen forcomplications and could potentially reduce themortality.

It is also important in the inpatient programs topromote at least a part of the program as a gender-based approach and to take into account specifi-cally male issues and needs, so as to define thera-peutic tools and techniques.

References

1. Lucas AR, Beard CM, O’Fallon WM, Kurland LT. 50-year trends

in the incidence of anorexia nervosa in Rochester, Minn.: A

population-based study. Am J Psychiatry 1991;148:917–922.

2. Garfinkel PE, Lin E, Goering P, Spegg C, Goldbloom DS, Kennedy

S, et al. Bulimia nervosa in a Canadian community sample:

Prevalence and comparison of subgroups. Am J Psychiatry

1995;152:1052–1058.

3. Braun DL, Sunday SR, Huang A, Halmi KA. More males

seek treatment for eating disorders. Int J Eat Disord 1999;25:

415–424.

4. Forman-Hoffman VL, Watson TL, Andersen AE. Eating disorder

age of onset in males: Distribution and associated characteris-

tics. Eat Weight Disord 2008;13:e28–e31.

5. Carlat DJ, Camargo CA Jr, Herzog DB. Eating disorders in males:

A report on 135 patients. Am J Psychiatry 1997;154:1127–1132.

6. Sharp CW, Clark SA, Dunan JR, Blackwood DH, Shapiro CM. Clin-

ical presentation of anorexia nervosa in males: 24 new cases.

Int J Eat Disord 1994;15:125–134.

7. Raevuori A, Hoek HW, Susser E, Kaprio J, Rissanen A, Keski-Rah-

konen A. Epidemiology of anorexia nervosa in men: A nation-

wide study of Finnish twins. PLoS One 2009;4:e4402.

8. Fernandez-Aranda F, Aitken A, Badıa A, Gimenez L, Solano R, Col-

lier D, et al. Personality and psychopathological traits of males

with an eating disorder. Eur Eat Disord Rev 2004;12:367–374.

9. Woodside DB, Kaplan AS. Day hospital treatment in males with

eating disorders—Response and comparison to females. J Psy-

chosom Res 1994;38:471–475.

10. Crisp A, Collaborators. Section 1: Seven Studies 1.5. Anorexia

nervosa in males: Similarities and differences to anorexia nerv-

osa in females. Eur Eat Disord Rev 2006;14:163–167.

11. Moller-Madsen S, Nystrup J, Nielsen S. Mortality in anorexia

nervosa in Denmark during the period 1970–1987. Acta Psy-

chiatr Scand 1996;94:454–459.

12. Burns T, Crisp AH. Factors affecting prognosis in male ano-

rexics. J Psychiatr Res 1985;19:323–328.

13. Crisp AH, Burns T, Bhat AV. Primary anorexia nervosa in the

male and female: A comparison of clinical features and prog-

nosis. Br J Med Psychol 1986;59 (Pt 2):123–132.

14. Fedorowicz VJ, Falissard B, Foulon C, Dardennes R, Divac SM,

Guelfi JD, et al. Factors associated with suicidal behaviors in a

large French sample of inpatients with eating disorders. Int J

Eat Disord 2007;40:589–595.

15. Huas C, Godart N, Foulon C, Pham-Scottez A, Divac S, Fedoro-

wicz V, et al. Predictors of dropout from inpatient treatment

for anorexia nervosa: Data from a large French sample. Psychi-

atry Res 2011;185:421–426.

16. Derogatis L. SCL-90 manual I: Administration, Scoring and Pro-

cedures Manual for the Revised Version. Baltimore, MD: Johns

Hopkins University School of Medicine, 1977.

17. Garner DM, Olmstead MP, Polivy J. Development and validation

of a multidimensional eating disorder inventory for anorexia

nervosa and bulimia. Int J Eat Disord 1983;2:15–34.

18. Garner DM, Garfinkel PE. The Eating Attitudes Test: An index of

the symptoms of anorexia nervosa. Psychol Med 1979;9:273–279.

19. Beck AT, Beamesderfer A. Assessment of depression: The

depression inventory. Mod Probl Pharmacopsychiatry

1974;7:151–169.

20. Huas C, Caille A, Godart N, Foulon C, Pham-Scottez A, Divac S,

et al. Factors predictive of ten-year mortality in severe anorexia

nervosa patients. Acta Psychiatr Scand 2011;123:62–70.

21. National Collaborating Centre for Mental Health. Eating Disor-

ders, Core interventions in the treatment and management of

anorexia nervosa, bulimia nervosa, and related eating disor-

ders-National Clinical Practice Guideline Number CG9: The Brit-

ish Psychological Society and Gaskell, 2004.

22. R Development Core Team. R: A language and environment for

statistical computing. Vienna, Austria: R Foundation for

Statistical Computing, 2004; Available from: http://www.

R-project.org.

23. Rothman KJ. No adjustments are needed for multiple compari-

sons. Epidemiology 1990;1:43–46.

24. Bouyer J, Hemon D, Cordier S, Derriennic F, Stucker I, Stengel

B, Clavel J. Epidemiologie. Principes et methodes quantitatives.

Inserm ed 2004.

25. Bland JM, Altman DG. The logrank test. BMJ 2004;328:1073.

26. Strober M, Freeman R, Lampert C, Diamond J, Teplinsky C,

DeAntonio M. Are there gender differences in core symptoms,

temperament, and short-term prospective outcome in anorexia

nervosa? Int J Eat Disord 2006;39:570–575.

27. Sepulveda AR, Carrobles JA, Gandarillas AM. Gender, school

and academic year differences among Spanish university stu-

dents at high-risk for developing an eating disorder: An epide-

miologic study. BMC Public Health 2008;8:102.

28. Rolland-Cachera MF, Cole TJ, Sempe M, Tichet J, Rossignol C,

Charraud A. Body mass index variations: Centiles from birth to

87 years. Eur J Clin Nutr 1991;45:13–21.

29. DRESS. Suicides et tentatives de suicides en France. Etudes

Result 2001;109:8.

30. Andersen AE. Diagnosis and tretment of males with eating dis-

orders. In: Andersen AE, editor. Males with Eating Disorders.

New York: Brunner/Mazel, 1990. pp.133–162.

31. Striegel-Moore RH, Garvin V, Dohm FA, Rosenheck RA. Eating

disorders in a national sample of hospitalized female and male

veterans: Detection rates and psychiatric comorbidity. Int J Eat

Disord 1999;25:405–414.

32. Tanner JM. Growth at Adolescence, 2nd ed. Oxford: Blackwell

Scientific Publications, 1962.

33. Ricciardelli LA, McCabe MP. A biopsychosocial model of disor-

dered eating and the pursuit of muscularity in adolescent boys.

Psychol Bull 2004;130:179–205.

34. Robb AS, Dadson MJ. Eating disorders in males. Child Adolesc

Psychiatr Clin N Am 2002;11:399–418, xi.

35. Carlat DJ, Camargo CA Jr. Review of bulimia nervosa in males.

Am J Psychiatry 1991;148:831–843.

GUEGUEN ET AL.


36. Chambry J, Agman G. L’anorexie mentale masculine a l’adoles-

cence. Psychiatr Enfant 2007;49:477–511.

37. Nielsen S, Moller-Madsen S, Isager T, Jorgensen J, Pagsberg K,

Theander S. Standardized mortality in eating disorders—A

quantitative summary of previously published and new evi-

dence. J Psychosom Res 1998;44:413–434.

38. Steinhausen HC. The outcome of anorexia nervosa in the 20th

century. Am J Psychiatry 2002;159:1284–1293.

39. Tanaka H, Kiriike N, Nagata T, Riku K. Outcome of severe ano-

rexia nervosa patients receiving inpatient treatment in Japan:

An 8-year follow-up study. Psychiatry Clin Neurosci

2001;55:389–396.

40. Deter HC, Schellberg D, Kopp W, Friederich HC, Herzog W. Pre-

dictability of a favorable outcome in anorexia nervosa. Eur Psy-

chiatry 2005;20:165–172.

41. Papadopoulos FC, Ekbom A, Brandt L, Ekselius L. Excess mortal-

ity, causes of death and prognostic factors in anorexia nervosa.

Br J Psychiatry 2009;194:10–17.

42. Zipfel S, Lowe B, Reas DL, Deter HC, Herzog W. Long-term prog-

nosis in anorexia nervosa: Lessons from a 21-year follow-up

study. Lancet 2000;355:721–722.

43. Keel PK, Dorer DJ, Eddy KT, Franko D, Charatan DL, Herzog DB.

Predictors of mortality in eating disorders. Arch Gen Psychiatry

2003;60:179–183.

44. Spillane NS, Boerner LM, Anderson KG, Smith GT. Comparability

of the eating disorder inventory-2 between women and men.

Assessment 2004;11:85–93.

45. Greenberg ST, Schoen EG. Males and eating disorders: Gender-

based therapy for eating disorder recovery. Prof Psychol Res

Pract 2008;39:464–471.

46. Oates-Johnson T, DeCourville N. Weight preoccupation, person-

ality, and depression in university students: An interactionist

perspective. J Clin Psychol 1999;55:1157–1166.

47. McCreary DR, Hildebrandt TB, Heinberg LJ, Boroughs M, Thomp-

son JK. A review of body image influences on men’s fitness goals

and supplement use. Am J Mens Health 2007;1:307–316.



Documents

Severe anorexia nervosa in men: Comparison with severe AN in women and analysis of mortality