Personality trait risk factors for attempted suicide among young women with eating disorders

European Psychiatry 19 (2004) 131–139 www.elsevier.com/locate/eurpsy

Original article

Personality trait risk factors for attempted suicide among young women with eating disorders G. Youssef a,*, B. Plancherel a, J. Laget a, M. Corcos b, M.F. Flament c, O. Halfon a a

Service Universitaire de Psychiatrie de l’Enfant et de l’Adolescent (Pr. O. Halfon), Lausanne, Switzerland b Département de Psychiatrie (Pr. Ph. Jeammet), Institut Mutualiste Montsouris, Paris, France c University of Ottawa and Royal Ottawa Hospital, Canada Received 4 December 2002; received in revised form 4 September 2003; accepted 21 November 2003 Available online 08 May 2004

Abstract Objective. – Clinical observations and a review of the literature led us to hypothesize that certain personality and character traits could provide improved understanding, and thus improved prevention, of suicidal behaviour among young women with eating disorders. Method. – The clinical group consisted of 152 women aged between 18 and 24 years, with DSM-IV anorexia nervosa/restrictive type (AN-R = 66), anorexia nervosa/purging type (AN-P = 37), bulimia nervosa/non-purging type (BN-NP = 9), or bulimia nervosa/purging type (BN-P = 40). The control group consisted of 140 subjects. The assessment measures were the Minnesota Multiphasic Personality Inventory—second version (MMPI-2) scales and subscales, the Beck Depression Inventory (BDI) used to control for current depressive symptoms, plus a specific questionnaire concerning suicide attempts. Results. – Suicide attempts were most frequent in subjects with purging behaviour (30.0% for BN-P and 29.7% for AN-P). Those attempting suicide among subjects with eating disorders were mostly students (67.8%). For women with AN-R the scales for ‘Depression’ and ‘Antisocial practices’ represented significant suicidal risk, for women with AN-P the scales for ‘Hysteria’, ‘Psychopathic deviate’, ‘Shyness/Self-consciousness’, ‘Antisocial Practices’, ‘Obsessiveness’ and ‘Low self-esteem’ were risk indicators and for women with BN-P the ‘Psychasthenia’, ‘Anger’ and ‘Fears’ scales were risk indicators. Conclusion. – This study provides interesting results concerning the personality traits of young women with both eating disorders and suicidal behaviour. Students and those with purging behaviour are most at risk. Young women should be given more attention with regard to the risk of suicide attempts if they: (a) have AN-R with a tendency to self-punishment and antisocial conduct, (b) have AN-P with multiple physical complaints, are not at ease in social situations and have antisocial behaviour, or (c) if they have BN-P and tend to be easily angered with obsessive behaviour and phobic worries. The MMPI-2 is an interesting assessment method for the study of traits indicating a risk of suicidal behaviour in young subjects, after controlling for current depressive pathology. © 2004 Elsevier SAS. All rights reserved. Keywords: Eating disorder; Suicide attempt; MMPI-2; Personality trait; Young adult; Adolescent

1. Introduction Growing concerns in recent years about suicidal behaviour have led to many studies aimed at understanding suicide risk factors and investigating the assessment and treatment of adolescents and young adults with suicidal behaviour. Attention was often given to those with co-morbid mental disorders [2,3], especially those with both a Diagnostic and Sta-

* Corresponding author. Unité d’Hospitalisation Psychiatrique pour Adolescents (UHPA), Centre Hospitalier Universitaire Vaudois (CHUV), 46 rue du Bugnon, BH-11, Lausanne 1011, Switzerland. E-mail address: [email protected] (G. Youssef). © 2004 Elsevier SAS. All rights reserved. doi:10.1016/j.eurpsy.2003.11.005

tistical Manual of Mental Disorders edition IV (DSM-IV) psychiatric diagnosis (Axis I) and personality disorders (Axis II) [5,7]. In contrast, the assessment of personality disorders in adolescence and young adulthood is difficult. Thus, efforts to measure individual characteristics using comprehensive measures of personality traits are essential. Very little is known about suicide among women with eating disorders, even though eating disorders substantially increase the risk of attempting suicide [3,38]. The mortality rate for anorexia nervosa is approximately 5.6% per decade, and suicide is the second most common cause of death after complications of the eating disorder [36]. Suicide attempts among subjects with bulimia nervosa seem to be linked to the presence of purging behaviour [16,17]. It also appears that

132

G. Youssef et al. / European Psychiatry 19 (2004) 131–139

depression, suicidal tendencies and anxiety are state dependent and in many cases resolve along with the remission of the bulimic symptoms [27]. A higher level of obsessive behaviour was found in both anorexic and bulimic suicide attempters compared to non-attempters, and in suicide repeaters compared to non-repeaters [17]. The mental-health disorders most frequently described as risk factors for suicide include substance-related disorders [37], personality disorders [5] and mood disorders [18,29], particularly depression [19,21,23,30]. For Stein et al. [35], subjects who were single and multiple suicide attempters demonstrated similarly high levels of depression, anxiety and impulsivity. For Goldston et al. [18], repeated suicide attempters reported more depressive symptoms. The elimination of affective disorders could markedly reduce the incidence of serious suicide attempts [2]. ‘Depression’ traits are often found in eating-disorder profiles [14,31]. Such traits should not be confused with depression defined as a mood disorder (DSM-IV), which could also be present among women with eating disorders. Many studies have indicated personality and character traits that are related to suicidal behaviour. Halfon et al. [20] showed that in a population of hospitalized adolescents with psychological problems, suicidal adolescents have more personality disorders. Brent et al. [7] found similar results. Beautrais et al. [4] showed a clear association between hopelessness, neuroticism and external locus of control and the risk of serious suicide attempts. Apter et al. [1] defined two types of suicidal behaviour: internalizing and externalizing. Kopper et al. [26] used the Minnesota Multiphasic Personality Inventory (MMPI) scales to describe profiles for suicide probability. Problem-solving skill deficit [41], lack of use of coping styles [22], deficit in self-esteem [19,40], low frustration tolerance [24], low social competence and psychosocial functioning [29,34], and problems with family and peer relationships [21,24] are recorded as suicide risk factors. Other studies have pointed out personality and character traits related to eating disorders [8,10,11,17,37,42]. Bulik et al. [9] described character traits among subgroups of eating disorders. Klump et al. [25] described temperament and character traits among anorexia nervosa subtypes. Schork et al. [32] showed that long-term, persistent anorexia nervosa was significantly correlated with some pathology scales of the MMPI that are elevated. Dancyger et al. [15] found similar results. Cumella et al. [14] showed that some MMPI-2 scales (Depression, Psychasthenia and Hysteria) and especially two-code profiles were more frequently present in eating disorders. Pryor and Wiederman [31] had similar results for scales and described code profiles for subgroups. Westen et al. [39] divided eating disorders into three personality-profile categories. Although much has been done to describe personality traits for suicidal behaviour and for eating disorders, little is known of personality trait risk factors for suicide attempts among those with eating disorders. The aim of this cross-sectional study is: (1) To verify the hypothesis derived from clinical experience that suicide risk among young women with eating

disorders is not only related to a simple addition of co-morbid factors (especially depression) leading to suicidal behaviour, but also to specific personality traits found in eating disorders. (2) To confirm the utility of the MMPI-2 scales and subscales (a widely used assessment method) in the follow-up of young women with eating disorders, to provide improved comprehension and evaluation of this population, and better prevention of the suicide risk that is present in this population. 2. Materials and methods 2.1. Population This cross sectional study is part of a larger research program on Addictive Disorders, from the Inserm (Institut National de la Santé et de la Recherche Médicale—France) clinical research programme (no. 494013-1994) 1. This project set out to identify psychopathological features common to all addictive behaviours, regardless of the nature of the addictive practice or behaviour (i.e. anorexia nervosa, bulimia nervosa, or alcohol and drug addiction). Data collection began in 1994 and was completed in 1998, while statistical analysis was conducted and the final research report was produced in 1999. From the 1593 files assembled, 102 (28 clinical subjects and 74 control subjects) were excluded: 26 did not meet inclusion criteria, two patients and 74 control subjects met one or more exclusion criteria (for description of criteria, see below). The final sample included 1491 subjects, 705 clinical subjects and 786 control subjects. The clinical subjects included 427 women and 278 men, and the control subjects included 469 women and 317 men. Clinical and control subjects were matched according to sex, age and socio-professional category. Matching followed the procedure of drawing lots in the Statistical Analysis System (SAS) software. The 705 clinical subjects included 660 with drug/alcohol-dependence or anorexia nervosa/bulimia nervosa. The remaining 45 were drug/alcohol abusers without the defined criteria. Some control subjects did not match with clinical subjects. Therefore, 615 control subjects finally matched with 660 clinical subjects. None of the control subjects was used to match both anorexia nervosa and bulimia nervosa groups at the same time. In the overall clinical group, the inclusion criteria were: (1) subjects asking for treatment in hospitals or in medical 1

The promoter is the “Institut Mutualiste Montsouris—Paris”, under the co-ordination of Professor Philippe Jeammet. The SUPEA (University Unit of Child and Adolescent Psychiatry—Professor Olivier Halfon) took part in this international multicentric research program (Belgium, France and Switzerland). As to the ethical aspects, the project was accepted by the CCPRB (Comité Consultatif de Protection dans la Recherche Biomédicale de Paris—Cochin) on the 24th of December 1995 (File No. 789), and the protocol was accepted by the CNIL (Commission Nationale de l’Informatique et des Libertés—France) on the 14th of February 1995 (File No. 364580).

G. Youssef et al. / European Psychiatry 19 (2004) 131–139

centres (in or outpatients), (2) subjects presenting with a current diagnosis of addictive disorder meeting the DSM-IV criteria, (3) inclusion in one of the two age groups evaluated (15–24-years-old and 25–45-years-old). The exclusion criteria were: (1) subjects with psychosis criteria as defined by the DSM-IV and (2) subjects not capable of correctly filling in the questionnaires. The evaluation was carried out during the first two weeks of hospitalisation and/or as soon as the patient’s general state allowed it (at a distance from the somatic complications for eating disorders and from acute intoxication for drug/alcohol dependent subjects). For eating disorders, the mean body mass index (BMI) values are shown in Table 1. A control group with no past or current addictive behaviour was constituted. The inclusion criteria were: (1) subjects between 15 and 45 years of age, (2) recruited among employees in medical centres, students from schools, colleges and universities, as well as subjects asking for treatment (in or outpatients) for minor medical or surgical problems. The exclusion criteria were: (1) subjects meeting psychosis criteTable 1 Mean body mass index (BMI) for suicide attempters and non-attempters in eating disorder subgroups Eating disorder subgroups AN-R AN-P BN-NP BN-P

NSA mean BMI

SA mean BMI (P for BMI)

14.78 15.24 23.02 20.28

14.15 (NS) 16.57 (NS) 23.52 (NS) 20.28 (NS)

NSA, no suicide attempts; SA, suicide attempters; BMI, body mass index; NS, not significant.

133

ria as defined by the DSM-IV, (2) subjects not capable of correctly filling in the questionnaires. Matching criteria were: (1) age (±1 year for 15–24-yearolds and ±3 years for 25–45-year-olds), (2) sex, and (3) socio-professional characteristics according to the National Institute of Statistics and Economical Studies in France (INSEE), [the INSEE distinguishes eight categories but we used a reconstitution into four categories (Categories 1 and 6: farmers and working-class; Categories 3 and 4: middle/ superior managers and superior intellectual professions; Categories 2 and 5: craftsmen/women, entrepreneur, company heads and employees; Category 8: other persons without professional activity)]. For the particular cases of students, these were matched with students (Table 2). For our cross-sectional study, we retrieved from the preceding clinical group female subjects aged 18–24 with eating disorders. We had an overall group of 160 subjects, from which we excluded eight subjects presenting past or current binge-eating disorder. We finally had 152 subjects divided into four clinical subgroups: Anorexia Nervosa/Restrictive type (AN-R = 66), Anorexia Nervosa/Purging type (ANP = 37), Bulimia Nervosa/Non-Purging type (BN-NP = 9) and Bulimia Nervosa/Purging type (BN-P = 40). The matching control sample consisted of 140 control subjects, 90 for the whole AN group and 50 for the whole BN group. We compared each of the ‘AN-subgroups’ to the AN control group (n = 90), and similarly each of the ‘BN-subgroups’ to the BN control group (n = 50). Other comparative criteria were subjects presenting with ‘at least one attempted suicide’ (SA) (Table 3) vs. subjects with ‘no attempted suicide’ (NSA). If there were significant differences in the frequency

Table 2 Socio-professional profile Socio-professional status Students Farmers and working class Managers and intellectuals Craftswomen and entrepreneur No professional activity

Eating disorder group Subjects Frequency (%) 103 67.8 1 0.7 31 20.4 9 5.9 8 5.3

Control group Subjects 94 1 29 9 7

Frequency (%) 67.1 0.7 20.7 6.4 5.0

Total

152

140

100.0

100.0

Table 3 Samples in eating disorder subgroups and their matching control groups Group Eating disorder

Control

Total AN AN-R AN-P BN BN-NP BN-P Total Matching AN Matching BN

SA, suicide attempters; S.D., standard deviation.

Total subjects (Mean age ± S.D.) 152 103 66 (20.29 ± 1.76) 37 (21.03 ± 1.83) 49 9 (21.44 ± 1.67) 40 (21.17 ± 1.69) 140 90 (20.48 ± 2.26) 50 (21.12 ± 2.02)

SA (Mean age/P for age) 36 22 11 (20.82/0.28) 11 (21.64/0.19) 14 2 (21.00/0.70) 12 (20.75/0.30) 9 6 (20.00/0.60) 3 (20.33/0.49)

Frequency (%) 23.7 21.4 16.7 29.7 28.6 22.2 30.0 6.4 6.7 6.0

134

G. Youssef et al. / European Psychiatry 19 (2004) 131–139

of suicide attempts between clinical groups and control groups [v2(1) = 18.38, P < 0.001], there was no significant difference in the frequency of suicide attempts between the four eating disorder subgroups [v2(3) = 3.62, P = 0.30]. 2.2. Assessment At initial presentation, all subjects were assessed by a psychiatrist or a clinical psychologist. The Mini International Neuropsychiaric Interview (MINI) [33], a short structured diagnostic interview which takes approximately 15 min to administer, provided most Axis I psychiatric diagnoses, while the DSM-IV criteria provided classification for subgroups of eating disorders. A suicide attempt was defined as a self-destructive act sufficiently serious to require medical evaluation, and carried out with the intent to end one’s life [28]. The suicide questionnaire included details of the date of the event, the method and severity of the suicide attempt. The severity of the attempt, relating to lethality, was assessed by the clinician as (a) minimal, (b) moderate, (c) serious, (d) extreme, or (e) unknown. In addition, all subjects completed the 13-item Beck Depression Inventory (BDI) [6], which assesses depressive symptoms (the higher the score, the more depressive the subject). The BDI refers to a time lapse before the assessment day, and characterizes the ‘subjective symptoms of depression’. All subjects also completed the MMPI-2™ [12], and scale and subscale scores were used to describe personality and character traits. They included

Clinical Scales, Clinical Subscales, Content Scales and Supplementary Scales (Table 4). The MMPI-2 consists of 567 true–false items and takes 60–90 min to complete. It gives a dimensional approach to the personality, i.e. a personality profile based on 10 scales, a symptomatic constellation and indices of the quality of interpersonal relationships and behavioural fragility. It proposes some diagnostic considerations and some therapeutic orientations. It is strict on the validity for a given patient, and it excludes patients who answer randomly or incorrectly. For the current study, we focused on scales which gave significant interactions with a history of SA (Tables 5–7). Analyses were performed on scales transformed to T-scores (mean = 50, standard deviation = 10). Higher scores reflect more serious states. Scores ≥ 65 are considered pathological. 2.3. Statistical analysis To compare the frequencies of suicide attempts (SA) between groups, a v2 test was used. To compare the means of the MMPI-2 scales and subscales between groups, an analysis of covariance (ANCOVA) was used, with BDI as covariate. For every MMPI-2 scale and subscale, an analysis of variance was performed by using the General Linear Model (GLM) procedure in the Statistical Package for the Social Sciences (Copyright SPSS Inc., 1989–2001). The main tested effects were: (1) history or no history of suicide attempt, (2) being in the clinical or in the control group, and (3)

Table 4 MMPI-2 scales and subscales Clinical scales Hypochondriasis Depression Hysteria Psychopathic deviate Masculinity– Femininity Paranoia Psychasthenia Schizophrenia Hypomania Social introversion

(Hs) (D) (Hy) (Pd)

Content scales Anxiety Fears Obsessiveness Depression

(ANX) (FRS) (OBS) (DEP)

(Mf)

Health concerns

(HEA)

(Pa) (Pt) (Sc) (Ma) (Si)

Bizarre mentation Anger Cynicism Antisocial practices Type A Low self-esteem Social discomfort Family problems Work interference Negative treatment indicators

(BIZ) (ANG) (CYN) (ASP) (TPA) (LES) (SOD) (FAM) (WRK) (TRT)

Supplementary scales Conscious anxiety Conscious repression Ego strength Mac Andrew Alcoholism—Revised Addiction potential Addiction admission Marital distress

(A) (R) (Es) (MAC-R)

Additional subscales Overcontrolled hostility Dominance Social responsibility College maladjustment

(O-H) (Do) (Re) (Mt)

(APS)

Masculine gender role

(GM)

(AAS) (MDS)

Feminine gender role PTSD-Kean PTSD-Schlenger Shyness/Self-consciousness Social avoidance Alienation—self and others

(GF) (PK) (PS) (Si 1) (Si 2) (Si 3)

Table 5 Analysis of variance for anorexia nervosa/restrictive type (significant results) MMPI-2 scales and subscales Depression Antisocial practices

AN-R (n = 66) Suicide attempts (n = 11) 62.89 57.04

No suicide attempts (n = 55) 60.66 52.53

CG (n = 90) Suicide attempts (n = 5) 60.56 44.40

SA, suicide attempt; BDI, Beck Depression Inventory scale; NS, not significant. * P < 0.05. ** P < 0.01. *** P < 0.001.

No suicide Attempts (n = 85) 54.42 52.50

Group effect

SA effect

Group × SA interaction

BDI control

NS NS

NS NS

* 0.0522

*** NS

G. Youssef et al. / European Psychiatry 19 (2004) 131–139

135

Table 6 Analysis of variance for anorexia nervosa/purging type (significant results) MMPI-2 scales and subscales

Hysteria Psychopathic deviate Shyness/Self-consciousness Antisocial practices Obsessiveness Low self-esteem

AN-P (n = 37) Suicide attempts (n = 11) 60.78 58.85 50.99 56.78 66.00 57.53

No suicide attempts (n = 26) 58.93 52.93 46.48 52.32 52.68 52.76

CG (n = 90) Suicide attempts (n = 5) 62.83 65.63 42.84 44.26 54.64 49.07

No suicide attempts (n = 85) 52.80 54.42 52.80 52.17 53.79 56.70

Group effect

SA effect

Group × SA interaction

BDI control

NS NS NS * NS NS

NS * NS NS * NS

* * ** * * NS

*** *** *** NS *** ***

SA, suicide attempt; BDI, Beck Depression Inventory scale, NS = not significant. * P < 0.05. ** P < 0.01. *** P < 0.001. Table 7 Analysis of variance for bulimia nervosa/purging type (significant results) MMPI-2 scales and subscales Psychasthenia Anger Fears

BN-P (n = 40) Suicide attempts (n = 12) 66.84 60.18 55.74

No suicide attempts (n = 38) 65.44 51.47 47.88

CG (n = 50) Suicide attempts (n = 3) 65.81 47.48 47.15

No suicide attempts (n = 47) 52.53 52.01 51.45

Group effect

SA Group × SA effect effect

BDI control

0.0580 NS NS

* NS NS

*** * **

0.0736 * 0.0613

SA, suicide attempt; BDI, Beck Depression Inventory scale; NS, not significant. * P < 0.05. ** P < 0.01. *** P < 0.001.

the interaction between (1) and (2). The major interest of this analysis was to identify significant interactions between these two factors. The alpha level was fixed at 0.05. Analyses were carried out for each of the eating disorder subgroups. Due to the low frequencies in some subgroups, the interaction test power needed for an acceptable effect size of at least 0.30 was 0.90. Thus, in the AN-R subgroup, the power was 0.90 for an effect size of 0.30, whereas in the AN-P subgroup, the power was 0.91 for the same effect size. On the other hand, in the BN-NP subgroup, a power of 0.85 was required to obtain an effect size of 0.40, and 0.96 to obtain 0.50. Finally, for the BN-P subgroup, an effect size of 0.40 gave a power of 0.96 (calculated using the G*Power program, by Buchner et al.; last update March 28, 2001). It should be noted that the MMPI-2 scales do not have the same meaning and significance as the DSM-IV psychiatric pathologies. MMPI-2 refers to ‘traits’ of character and personality, while DSM-IV refers to ‘symptoms’ and psychiatric diagnoses. It should thus be specified that the scales in the MMPI-2 do not measure ‘Depression’ with the same meaning as in the psychiatric diagnostic classification of the DSMIV. The term ‘depression’ could lead to confusion. Therefore, ‘Depression’ will be referred to as a personality trait (MMPI2), while ‘depression’ will be referred to as a symptom (DSM-IV). The trait ‘Depression’ reflects not only feelings of discouragement, pessimism, or despair, but also fundamental traits of personality such as excessive sense of responsibility, exaggerated self-demands and a tendency to self-punishment. The same arguments apply for other scales but those scales lead to less confusion and can be more easily identified. In the statistical procedures, we compared subgroups after controlling for the depression measured by the

BDI, because ‘Depression’ (in terms of diagnosis) is an important suicide risk factor (see above). BDI scores showed a very strong correlation with most of the MMPI-2 scales. Excluding ‘Depression’ symptoms enabled us to study the MMPI-2 traits without interference. We will therefore give estimated mean values using the GLM, after the introduction of the BDI co-variate (in addition to the two other independent variables, Group and Suicide Attempt). Thus, corrected mean values were studied.

3. Results Relating to socio-professional status (Table 2), most eating disorder subjects were students (67.8%) or managers/ intellectuals (20.4%), followed by craftswomen/entrepreneur classes (5.9%), no professional activity (5.3%), and finally farmers and working classes (0.7%). Mean age was approximately 21 years (Table 3). Socio-professional activities were similarly distributed in both eating disorder and control groups (Table 2), even though the total number of subjects in each group was not equal. Mean age values for purging groups (AN-P and BN-P) were slightly higher than mean age values for the restrictive group (AN-R). Mean age values for suicide attempters in the eating disorder subgroups were almost the same as in the control groups, and there was no statistically significant difference between the mean age values of SA and the mean age values of NSA in the different subgroups of eating disorders and control groups. In the whole eating disorder group, 23.7% had attempted suicide, compared to only 6.4% in the whole control group. Most suicide attempters were in the purging groups, AN-P

136

G. Youssef et al. / European Psychiatry 19 (2004) 131–139

(29.7%) and BN-P (30.0%). The mean BMI were not statistically significantly different for SA compared to NSA subjects in the various eating disorder subgroups (Table 1). The frequency of SA among bulimia nervosa women (28.6%) was higher than that in women with anorexia nervosa (21.4%) (Table 3). Not all MMPI-2 scales and subscales showed significant group differences. Interactions were sought for all scales and subscales (Table 4), but only those with significant interactions will be mentioned (Tables 5–7). In the AN-R subgroup (Table 5), the mean values of the ‘Depression’ and ‘Antisocial practices’ scales were higher for AN-R with suicide attempts than for AN-R without suicide attempts. The mean value of ‘Antisocial practices’ for the control group (CG) with suicide attempts was lower than the mean value for the CG without suicide attempts, while the mean value of ‘Depression’ for CG with suicide attempts was higher than the mean value for CG without suicide attempts. The Group–Suicide Attempt interaction was significant for ‘Depression’ after controlling for BDI score. The Group– Suicide Attempt interaction was slightly significant for ‘Antisocial practices’ but not when controlling for BDI score. The difference between the mean values of ‘Depression’ for AN-R subjects with and without suicide attempts was lower than the difference between the mean values for CG subjects with and without suicide attempts. In the AN-P subgroup (Table 6), the mean values of the scales for ‘Hysteria’, ‘Psychopathic deviate’, ‘Shyness/Selfconsciousness’, ‘Antisocial practices’, ‘Obsessiveness’ and ‘Low self-esteem’ were higher for AN-P with suicide attempts than for AN-P without suicide attempts. However, the mean values of ‘Shyness/Self-consciousness’, ‘Antisocial practices’ and ‘Low self-esteem’ for CG with suicide attempts were lower than the mean values for CG without suicide attempts. After controlling for BDI score, the Group– Suicide Attempt interactions were significant for ‘Hysteria’, ‘Psychopathic deviate’, ‘Shyness/Self-consciousness’, ‘Antisocial practices’, ‘Obsessiveness’ and ‘Low self-esteem’, but not for the ‘Antisocial practices’ scales, while it was significant for ‘Low self-esteem’ even though its Group– Suicide Attempt interaction was slightly significant. Mean values of ‘Obsessiveness’ for CG with and without suicide attempts were almost equal, while the mean value for AN-P with suicide attempts was much higher than the mean value for AN-P without suicide attempts. While mean values of ‘Hysteria’ and ‘Psychopathic deviate’ were higher for suicide attempters than for non-suicide attempters in both groups (AN-P and CG), mean values of ‘Hysteria’ and ‘Psychopathic deviate’ for AN-P with suicide attempts were lower than the mean values for CG with suicide attempts. For bulimia nervosa, the BN-NP subgroup sample was very small. The scale mean values did not give important statistical information, but this subgroup had a 22.2% prevalence of suicide attempts (Table 3). In the BN-P subgroup (Table 7), the mean values of ‘Psychasthenia’, ‘Anger’ and ‘Fears’ were higher for BN-P with suicide attempts than for the BN-P without suicide attempt. However, the mean values

of ‘Anger’ and ‘Fears’ for CG with suicide attempts were lower than those for CG without suicide attempt. After controlling for BDI score, the Group–Suicide Attempt interaction was significant for ‘Anger’, ‘Psychasthenia’ and ‘Fears’.

4. Discussion Even though this study did not have one-to-one matching criteria, it can generally be observed that both the anorexia nervosa group and the bulimia nervosa group and their matching control groups had similar mean age values, and that the distribution of various socio-professional activities was almost identical between the eating disorder groups and the control groups. These factors show that our study provides a close comparison between clinical subjects and control subjects, even though the latter group was chosen from a large variety of individuals (see above). From a general point of view, and according to our results, eating disorders mostly affect students, especially those aged around 21 years. Age in itself and BMI were not risk factors for suicide attempts in our study (Tables 1 and 3), unlike in the study by Favaro and Santonastaso [17], who found that anorexia nervosa suicide attempters were older with a lower BMI (in a population aged 12–48 years). As in the literature [16,17], suicide attempters were more often found among subjects with ‘purging’ behaviour. Suicide attempts were also more frequent among women with bulimia nervosa in general, results also comparable to the literature [16]. Moreover, our results on the prevalence of suicide attempts among women with bulimia nervosa (28.6%) were close to those found by Corcos et al. [13] (27.8%). Depression, as a psychiatric pathology, was often shown to be a suicide risk factor. Some studies have described depression as state dependent in eating disorders [27], others have described it as a co-morbid risk factor [3], and yet others have described it as a trait in eating disorders [14,31]. Our results show strong correlations between the BDI score and most of the MMPI-2 scale and subscale scores, indicating the necessity of excluding depressive symptoms before studying traits as risk factors. We therefore controlled subjects for depressive symptoms to exclude the influence of psychiatric co-morbidity, and obtained interesting results. In the AN-R subgroup, ‘Depression’ and ‘Antisocial practices’ appear as suicide risk factors. For ‘Depression’, scores are not much higher in AN-R with suicide attempts than in AN-R without suicide attempts, and ‘Depression’ influences the CG more in terms of suicide attempts. This may be explained by the fact that AN-R often have ‘Depression’, as described in the literature [14,31]. While depression (as a psychiatric pathology) is also often clinically present in eating disorders and is described as a risk factor for suicide, Bulik et al. [10] found that the presence of depression is not related to risk of suicide attempts in anorexia nervosa. We found that the MMPI trait of ‘Depression’ had more impact in the CG, once depressive symptoms were excluded. ‘Antiso-

G. Youssef et al. / European Psychiatry 19 (2004) 131–139

cial practices’ is a serious factor. More subjects have this trait in AN-R with suicide attempts than in AN-R without suicide attempts. On the other hand, fewer subjects have this trait in the CG with suicide attempts than in the CG without suicide attempts. This trait is almost equally present in the CG without suicide attempts and in the AN-R without suicide attempts. Moreover, there was only a low correlation between the BDI and the MMPI-2 concerning this trait, suggesting that having the ‘Antisocial practices’ trait, with or without the presence of depressive symptoms, is an important suicidal risk factor. In the AN-P subgroup, ‘Hysteria’, ‘Psychopathic deviate’, ‘Shyness/self-consciousness’, ‘Antisocial practices’, ‘Obsessiveness’ and ‘Low self-esteem’ are suicide risk factors. The difference between subjects with ‘Hysteria’ in the CG with suicide attempts and those in the CG without suicide attempts is higher compared to the difference in the AN-P subgroup. Therefore, ‘Hysteria’ has more influence in controls in terms of suicide attempts. Moreover, scores are not much higher in the AN-P with suicide attempts than in the AN-P without suicide attempts. This may be explained by the fact that the subjects in the AN-P group generally have high scores for ‘Hysteria’, as described in the literature [14,31]. ‘Shyness/self-consciousness’, ‘Antisocial practices’ and ‘Low self-esteem’ appear as more significant factors. More subjects have those traits in the AN-P with suicide attempts than in AN-P without suicide attempts. On the other hand, fewer subjects have this trait in the CG with suicide attempt than in CG without suicide attempts. ‘Shyness/Selfconsciousness’ is one of the Social Introversion subscales. Social alienation and poor social adjustment [32] have been described in eating disorders. The trait ‘Antisocial practices’ is almost equally present in CG without suicide attempts and in AN-R without suicide attempts. There was no significant correlation between BDI and MMPI-2 concerning this trait, suggesting that having the ‘Antisocial practices’ trait, with or without the presence of depressive symptoms, is an important suicide risk factor. The presence of eating disturbances and aggressive behaviour together has already been recorded as a profile representing a greater risk of attempting suicide [37]. Findings in the literature [19,40] and our results are similar in relation to the link between the deficit in selfesteem and the risk of suicide. Finally, ‘Obsessiveness’ is found almost equally in CG with and without suicide attempts, while it has far greater incidence in AN-P with suicide attempts than in AN-P without suicide attempts, suggesting that it is also an important risk factor for AN-P. These results are similar to those in the literature [5,8,17]. In the BN-NP subgroup, our clinical sample was not large enough to study MMPI-2 scale interactions. This group is the least commonly found in our clinical practice. However, the prevalence of suicide attempts (22.2%) was close to that given by Corcos et al. [13], who had a larger BN-NP subgroup (22%). In the BN-P subgroup, ‘Psychasthenia’, ‘Anger’ and ‘Fears’ are suicide risk factors, and ‘Anger’ and ‘Fears’ are

137

the most serious. More subjects amongst the BN-P with suicide attempts have these traits than subjects amongst the BN-P without suicide attempts, and in contrast, fewer subjects have this trait in the CG with suicide attempts than in the CG without suicide attempts. Subjects with ‘Anger’ traits tend to be irritable and stubborn, to lose control easily and may be physically violent. For Thompson et al. [37], the presence of both eating disorders and aggressive behaviour indicated a higher risk of suicide attempt. ‘Psychasthenia’ appears a much more serious factor for CG than for BN-P, given that the results are almost equivalent for BN-P with suicide attempts, BN-P without suicide attempts and CG with suicide attempts. This may be explained by the fact that this trait is generally found in eating-disorder profiles [14,31]. It includes obsessive ideation, compulsive rituals and exaggerated fears, and reflects general anxiety and distress. ‘Fears’ indicates that a subject has particular phobias. A potential limitation of our study is its cross sectional design. We have focused on eating disorders, while the larger initial study (on which ours was based) aimed to study a variety of dependence behaviours. Moreover, the initial study was not only focused on suicidal behaviour, but also on a wide range of psychopathological dimensions. It would have been interesting to get more specific information on the current topic, such as the number of previous suicide attempts, and their influence on suicidal behaviour [17,18]. However, for this, larger samples would have been needed, based primarily on eating disorders and suicide attempts. Furthermore, one-to-one matching criteria would have given more precise comparisons between eating disorders and control groups. Finally, the BDI is aimed to assess current depressive states, while eating disorders can be chronic pathologies, and it is not possible to infer past from current association between the two conditions. Only prospective longitudinal studies could give more precise information regarding the relationship between the course of eating disorders (associated with specific individual personality traits), depression and suicidal behaviour. However, the exclusion of the major mental health disorders described in the literature as risk factors for suicidal behaviours, such as substancerelated disorders and mood disorders, especially depression, as well as somatic complications of the eating disorders, allowed us to obtain results pertaining to trait rather than state characteristics. In addition, the MMPI-2 provides important information about personality traits in an age range where it is difficult to study personality structures using a categorical approach. In conclusion, among women with eating disorders, suicide attempts are most frequent among those with ‘purging’ behaviours. The AN-R group has the lowest percentage of suicide attempts. BN-NP is the least common eating disorder. As this clinical subgroup was small, results were not statistically significant. However, in general, subjects with bulimia nervosa have a higher risk of suicide attempt than those with anorexia nervosa. Age does not seem to have an influence on the risk of suicide attempt in the present study, although

138

G. Youssef et al. / European Psychiatry 19 (2004) 131–139

students are most at risk of eating disorders, with the mean age of our samples being around 21 years. The eatingdisorder subgroups each show specific personality traits associated with the risk of suicide attempts. The suicidal AN-R subjects tend to be easily discouraged, pessimistic and desperate, with an excessive sense of responsibility, exaggerated self-demands and a tendency towards self-punishment. They also have misanthropic attitudes and they present antisocial conducts. The AN-P subjects with a history of suicide attempt tend to deny problems, have an important need for affection, and frequently express lassitude and discomfort, with multiple physical complaints and inhibition of aggressiveness. They also have regular problems with authority, difficulties in decision-taking, and a tendency to excessive rumination and worry. Moreover, they tend to be shy, easily embarrassed, and not at ease in social, especially new situations; they have a negative opinion of themselves and lack self-confidence. Finally, the BN-P suicidal subjects tend to have obsessive ideation and compulsive rituals, as well as rigorous moral principles, an excessive sense of responsibility and self-accusation, making important efforts in a rigid way to control their impulsivity. Moreover, they have difficulties in controlling their anger and easily lose their selfcontrol; they have numerous fears and specific phobias. Our study confirms the hypothesis that women with eating disorders have specific personality traits associated with a suicidal risk, independently of the presence of current depressive symptoms. The study provides interesting results concerning the personality traits of young women with both eating disorders and suicidal behaviour. Students and those with purging behaviour are most at risk. Young women should be given more attention with regard to the risk of suicide attempts if they: (a) have AN-R with a tendency to self-punishment and antisocial conduct, (b) have AN-P with multiple physical complaints, are not at ease in social situations and have antisocial behaviour, or (c) if they have BN-P and tend to be easily angered with obsessive behaviour and phobic worries. The MMPI-2 is an interesting assessment method for the study of traits indicating a risk of suicidal behaviour among young subjects, after controlling for current depressive pathology, thus offering improved evaluation and follow up.

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

Acknowledgements We wish to thank Professor P. Baumann for his advice in finalizing this article.

References [1]

Apter A, Gothelf D, Orbach I, et al. Correlation of suicidal and violent behavior in different diagnostic categories in hospitalized adolescent patients. Journal of the American Academy of Child and Adolescent Psychiatry 1995;34:912–8.

[19]

[20]

[21]

Beautrais AL, Joyce PR, Mulder RT, Fergusson DM, Deavoll BJ, Nightingale SK. Prevalence and comorbidity of mental disorders in persons making serious suicide attempts: a case-control study. American Journal of Psychiatry 1996;153(8):1009–14. Beautrais AL, Joyce PR, Mulder RT. Psychiatric illness in a New Zealand sample of young people making serious suicide attempts. New Zealand Medical Journal 1998;111:44–8. Beautrais AL, Joyce PR, Mulder RT. Personality traits and cognitive styles as risk factors for serious suicide attempts among young people. Suicide and Life-Threatening Behavior 1999;29(1):37–47. Beautrais AL. Risk factors for suicide and attempted suicide among young people. Australian and New Zealand Journal of Psychiatry 2000;34:420–36. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuring depression. Archives of General Psychiatry 1961;4: 561–71. Brent DA, Johnson B, Bartle S, Rather C, Matta J, Connolly J, et al. Personality disorder, tendency to impulsive violence, and suicidal behaviour in adolescents. Journal of the American Academy of Child and Adolescent Psychiatry 1993;32(1):69–75. Bulik CM, Beidel DC, Duchmann E, Weltzin TE. Comparative psychopathology of women with bulimia nervosa and obsessivecompulsive disorder. Comprehensive Psychiatry 1992;33(4):262–8. Bulik CM, Sullivan PF, Weltzin TE, Kaye WH. Temperament in eating disorders. International Journal of Eating Disorders 1995;17(3):251–61. Bulik CM, Sullivan PF, Joyce PR. Temperament, character and suicide attempts in anorexia nervosa, bulimia nervosa and major depression. Acta Psychiatrica Scandinavia 1999;100(1):27–32. Bulik CM, Sullivan PF, Fear JL, Pickering A. Outcome of anorexia nervosa: eating attitudes, personality, and parental bonding. International Journal of Eating Disorders 2000;28(2):139–47. Butcher JN, Megargee EI. In: Hathaway SR, McKinley JC, editors. Minnesota Multiple Personality Inventory-2™. University of Minnesota Press; 1989 [Revision of the original MMPI®]:4. Corcos M, Taieb O, Benoit-Lamy S, Jeammet P, Flament MF. Suicide attempts in women with bulimia nervosa: frequency and characteristics. Acta Psychiatrica Scandinavia 2002;106:381–6. Cumella EJ, Wall AD, Kerr-Almeida N. MMPI-2 in the inpatient assessment of women with eating disorders. Journal of Personality Assessment 2000;75(3):387–403. Dancyger IF, Sunday SR, Eckert ED, Halmi KA. A comparative analysis of Minnesota Multiphasic Personality Inventory profiles of anorexia nervosa at hospital admission, discharge, and 10-year follow-up. Comprehensive Psychiatry 1997;38(3):185–91. Favaro A, Santonastaso P. Purging behaviors, suicide attempts, and psychiatric symptoms in 398 eating disordered subjects. International Journal of Eating Disorders 1996;20(1):99–103. Favaro A, Santonastaso P. Suicidality in eating disorders: clinical and psychological correlates. Acta Psychiatrica Scandinavia 1997;95(6): 508–14. Goldston DB, Daniel S, Reboussin DM, Kelley A, Ievers C, Brunstetter R. First-time suicide attempters, and previous attempters on an adolescent inpatient psychiatry unit. Journal of the American Academy of Child and Adolescent Psychiatry 1996;35(5):631–9. Groholt B, Ekeberg O, Wichstrom L, Haldorsen T. Young suicide attempters: a comparison between a clinical and an epidemiological sample. Journal of the American Academy of Child and Adolescent Psychiatry 2000;39(7):868–75. Halfon O, Laget J, Barrie M. An epidemiological approach to adolescent suicide. A comparison between suicidal and non-suicidal clinical groups in a health foundation center for French students. European Child and Adolescent Psychiatry 1995;4:32–8. Hollis C. Depression, family environment, and adolescent suicidal behavior. Journal of the American Academy of Child and Adolescent Psychiatry 1996;35(5):622–30.

G. Youssef et al. / European Psychiatry 19 (2004) 131–139 [22] Horesh N, Rolnick T, Iancu I, Dannon P, Lepkifker E, Apter A, et al. Coping styles and suicide risk. Acta Psychiatrica Scandinavia 1996; 93(6):489–93. [23] Houston K, Hawton K, Shepperd R. Suicide in young people aged 15–24: a psychological autopsy study. Journal of Affective Disorders 2001;63:159–70. [24] Kienhorst ICWM, de Wilde EJ, Diekstra RFW, Wolters WHG. Adolescents’ image of their suicide attempt. Journal of the American Academy of Child and Adolescent Psychiatry 1995;34(5):623–8. [25] Klump KL, Bulik CM, Pollice C, et al. Temperament and character in women with anorexia nervosa. Journal of Nervous and Mental Disease 2000;188(9):559–67. [26] Kopper BA, Osman A, Osman JR, Hoffman J. Clinical utility of the MMPI-A and Harris-Lingoes subscales in the assessment of suicidal risk factors in psychiatric adolescents. Journal of Clinical Psychology 1998;54(2):191–200. [27] Lehoux PM, Steiger H, Jabalpurlawa S. State/trait distinction in bulimic syndromes. International Journal of Eating Disorders 2000; 27(1):36–42. [28] Mann J, Waternaux C, Haas GL, Malone KM. Toward a clinical model of suicidal behaviour in psychiatric patients. American Journal of Psychiatry 1999;156(2):181–9. [29] Pelkonen M, Marttunen M, Pulkkinen E, Laippala P. Characteristics of out-patient adolescents with suicidal tendencies. Acta Psychiatrica Scandinavia 1997;95(2):100–7. [30] Pinto A, Whisman MA. Negative affect and cognitive biases in suicidal and non suicidal hospitalized adolescents. Journal of the American Academy of Child and Adolescent Psychiatry 1996;35(2):158– 65. [31] Pryor T, Wiederman MW. Use of the MMPI-2 in the outpatient assessment of women with Anorexia Nervosa or Bulimia Nervosa. Journal of Personality Assessment 1996;66(2):363–73. [32] Schork EJ, Eckert ED, Halmi KA. The relationship between psychopathology, eating disorder diagnosis, and clinical outcome at 10-year follow-up in anorexia nervosa. Comprehensive Psychiatry 1994;35(2):113–23.

139

[33] Sheehan DV, Lecrubier Y, Sheehan KH, Amorin P, Janavs J, Weiller E, et al. The Mini-International Neuropsychiatric Interview (MINI): the development and validation of structured diagnostic psychiatric interview for DSM-IV and ICD-10. Journal of Clinical Psychiatry 1998;59(Suppl 20):22–33. [34] Sourander A, Helstela L, Haavisto A, et al. Suicidal thoughts and attempts among adolescents: a longitudinal 8-year follow-up study. Journal of Affective Disorders 2001;63(1–3):241–7. [35] Stein D, Apter A, Ratzoni G, Har-Even D, Avidan G. Association between multiple suicide attempts and negative affects in adolescents. Journal of the American Academy of Child and Adolescent Psychiatry 1998;37(5):488–94. [36] Sullivan PF. Mortality in anorexia nervosa. American Journal of Psychiatry 1995;152:1073–4. [37] Thompson KM, Wonderlich SA, Crosby RD, Mitchell JE. The neglected link between eating disturbances and aggressive behavior in girls. Journal of the American Academy of Child and Adolescent Psychiatry 1999;38(10):1277–84. [38] Warshaw MG, Massion AO, Peterson LG, Pratt LA, Keller MB. Suicidal behavior in patients with panic disorder: retrospective and prospective data. Journal of Affective Disorders 1995;34(3):235–47. [39] Westen D, Harnden-Fischer J. Personality profiles in eating disorders: rethinking the distinction between Axis I and Axis II. American Journal of Psychiatry 2001;158:547–62. [40] Wichstrom L. Predictors of adolescent suicide attempts: a nationally representative longitudinal study of Norwegian adolescents. Journal of the American Academy of Child and Adolescent Psychiatry 2000; 39(5):603–10. [41] Wilson KG, Stelzer J, Bergman JN, Kral MJ, Inayatullah M, Elliott CA. Problem solving, stress, and coping in adolescent suicide attempts. Suicide and Life Threatening Behavior 1995;25:241–52. [42] Yamaguchi N, Kobayashi J, Tachikawa H, Sato S, Hori M, Suzuki T, et al. Parental representation in eating disorder patients with suicide. Journal of Psychosomatic Research 2000;49(2):131–6.