Eur Child Adolesc Psychiatry (2008) 17:392396 DOI 10.

1007/s00787-008-0681-0

ORIGINAL CONTRIBUTION

Carlo Marchesi Paolo Ampollini Chiara DePanlis Carlo Maggini

Temperament features in adolescents with ego-syntonic or ego-dystonic obsessive-compulsive symptoms

Accepted: 27 December 2007 Published online: 21 April 2008

j Abstract The present study

C. Marchesi, MD C. DePanlis, MD C. Maggini, MD Department of Neuroscience Psychiatric Division University of Parma Parma, Italy P. Ampollini, MD Mental Health Department AUSL Parma Parma, Italy C. Marchesi, MD (&) Dipartimento di Neuroscienze Sezione di Psichiatria ` Universita degli Studi di Parma Str del Quartiere 2 43100 Parma, Italy Tel.: +39-0521/703523 Fax: +39-0521/230611 E-Mail: carlo.marchesi@unipr.it

evaluated whether different patterns of temperament may predict a different threshold of acceptability of obsessive-compulsive (OC) symptoms in adolescents. OC symptomatology was detected with the Leyton Obsessional Inventory-Child Version (LOI-CV) and temperament was assessed using the tridimensional personality questionnaire in 2,775 highschool students. According to the LOI-CV scores, the adolescents were classied as high interference (interfering, ego-dystonic symptoms) (HI), supernormal (noninterfering, ego-syntonic symptoms) (Sn) and controls (C) HI were 119 (4.3%), Sn 85 (3.1%) and C 2,571 (92.6%). The best predictor of belonging to HI or Sn groups was the temperament conguration of high Harm Avoidance (HA) and high Persistence (P). The feature that mainly distinguishes the two

symptomatic groups were Novelty Seeking (NS) levels. Our data suggest that people characterized by pessimistic worry in anticipation of future problems, passive avoidant behaviour, rapid fatigability (high HA) and irresoluteness, ambitiousness, perseverance, perfectionism, enduring feelings of frustration (high P) might develop OC symptoms. Whether OC symptoms become ego-syntonic or ego-dystonic seems to mainly depend on NS levels: low NS might protect people (with the prevention of exploratory and active behaviours that may elicit loss of control on symptoms) from the development of interfering OC symptoms.
j Key words obsessive-compulsive disorder obsessive-compulsive personality disorder temperament adolescents

Introduction
Obsessive-compulsive disorder (OCD) is a serious mental illness, with onset occurring in childhood or adolescence in at least half of all cases [23]. Nevertheless very low attention has been given to childhood and adolescent OCD compared to the adult disorder. Obsessive-compulsive (OC) symptoms are a very common experience [21] but individual variations

seem to be important to establish the threshold of acceptability of these phenomena. In fact, in most cases thoughts and rituals are easily dismissed, whereas in others they reach a very high level of interference in the everyday life. In a study concerning the prevalence of adolescent OCD in a population of about 5.000 high school students, Flament et al [10] described two distinct groups of subject. The rst group, named High Interference (HI), included subjects with high interfering

ECAP 681

C. Marchesi et al. Temperament and obsessive-compulsive symptoms in adolescents

393

OC symptoms, suggesting the presence of clinical OCD. The second group, called Supernormals (Sn), consisted of subjects that are considered healthy or even supernormal individuals with noninterfering, ego-syntonic obsessional features. The same methodology and subgrouping were used in other two studies in Denmark [25] and in Italy by our research group [18]. The prevalence of each group differ among studies, ranging from 1.7 to 4.1% and 0.7 to 3.0% for HI and Sn groups, respectively. The possible explanations for these differences are discussed elsewhere [18]. However, none of these studies investigated whether HI and Sn subjects could be distinguished on the basis of features (biological markers, familiarity, personality, etc), which may help to understand the mechanisms involved in raising or lowering the threshold of acceptability of obsessions and compulsions and in determining the degree of interference in daily life activities. One of the most intriguing hypothesis is that temperament and personality features could represent risk factors for the development of specic mental disorders [8] and could therefore account for the differences in the threshold of acceptability of OC symptomatology. Cloninger and colleagues [4, 5, 9] developed a testable model of personality that provides a framework to describe and assess both the genetically inherited tendencies of temperament and the developmentally dependent dimensions of character, reecting personality development in the context of insight learning and environmental experiences. Particularly, the temperament seems to be more closely related to the risk of developing specic forms of mental disorders because of its biological implications [9]. Therefore, in the present study we evaluated the temperament features and OC symptoms in a large group of Italian High School students. The aim of the study was to test the hypothesis that different patterns of temperament may predict a different threshold of acceptability of obsessive-compulsive phenomena.

An informed written consent was obtained after the investigation was fully described to students, parents, and teachers.

j Assessment
The study was carried out by the mean of selfadministered questionnaires to be completed within a period of about 100 min. Detailed instructions were given by one of the investigators (AP) who remained in the classroom for the entire duration of the trial to avoid misunderstanding of questions. Obsessive-compulsive symptomatology was detected according to the italian version of the Leyton Obsessional Inventory-Child Version (LOI-CV) [3]. We chose this questionnaire because it has been proven to be a valid screening instrument and because it allowed us to compare our data with those of previous studies performed in the USA [10] and in Denmark [25]. This instrument consists of a 20-item questionnaire asking for the presence or absence of a number of obsessive preoccupations and behaviours (Yes Score). Furthermore, for each positive response, a rating of interference in personal functioning (range 03; no interference to interferes a lot) (Interference Score) were measured. According to the indications of Flament et al [10], two distinct groups of subjects were detected. The rst group, named High Interference (HI), included all the subjects scoring 25 or more on Interference Score, regardless of the Yes Score. In fact, the Interference Score, following Flament et als indications, had been the best indicator of psychopathology. The second group, called Supernormals (Sn), consisted of all subjects with a Yes Score of 15 or more and an Interference Score of 10 or less. These subjects are considered healthy or even supernormal individuals with noninterfering, ego-syntonic obsessional features. Subjects, with a Yes score lower than 15 at the LOI-CV and an Interference Score lower than 10, were classied as controls (C). Temperament was assessed with the Tridimensional Personality Questionnaire (TPQ) [6]. The TPQ was rated using the nal revision who allows the calculation of 4 temperament dimensions: Novelty Seeking (NS), Reward Dependence (RD), Persistence (P), and Harm Avoidance (HA).

Methods
j Sample
This study is part of an epidemiological investigation evaluating the prevalence of many psychiatric disorders in a sample of about 3,000 high-school students, attending the last 2 years of ten state high-schools in Parma (Italy). Details of the main study are given elsewhere [17, 18].

j Statistical analysis
The differences among HI, Sn and C groups (independent variable) in temperament dimensions scores (dependent variables) were evaluated with the one-

394

European Child & Adolescent Psychiatry (2008) Vol. 17, No. 6 Steinkopff Verlag 2008

Table 1 Temperament dimensions in high interference (HI), supernormal (Sn) and control (C) adolescents Temperament dimensions High interference n 119 Novelty seeking Harm avoidance Reward dependence Persistence 18.33 20.61 14.11 4.94 5.5 6.0 4.7 2.0 Supernormal n 85 16.48 18.48 14.32 5.27 5.1 5.7 4.1 2.0 Control n 2,571 18.04 15.89 13.18 4.42 5.2 5.5 4.5 2.0

two years of ten state high-schools in Parma (Italy), and completing the LOI-CV and the TPQ. One-hundred nineteen adolescents (4.3%) were classied as HI and 85 (3.1%) as Sn. The remaining 2,571 subjects (92.6%) were classied as C. Female gender was signicantly more frequent in HI (n 78; 65.5%) and in Sn (n 55; 64.7%) groups than in C group (n 1,249; 48.6%) (v2 = 20.8; df = 2; P < 0.001).

j Temperament dimensions (Table 1)

The levels of HA were higher in HI than in Sn subjects and in Sn than in C subjects. The levels of NS were lower in Sn than in HI and C groups (without signicant difference between HI and C subjects), and RD and P levels were higher in HI and Sn than in C (without differences between HI and Sn). After controlling for the effect of gender, the difference in RD scores disappeared.

One-way ANOVA, with Bonferroni post hoc analysis (df 2; 2,772) Novelty seeking: F = 4.0; P = 0.018 (Sn < HI, C ); Harm avoidance: F = 48.7; P < 0.001 (HI > Sn > C); Reward dependence: F = 4.9; P = 0.007 (Sn > C); Persistence: F = 10.8; P < 0.001 (Sn, HI > C); Analysis of covariance (covariate: gender) (df = 3; 2,775); Novelty seeking: F = 5.1; P = 0.006 (HI, C > Sn); Harm avoidance: F = 41.2; P < 0.001 (HI > Sn > C); Reward dependence: F = 1.0; P = 0.351; Persistence: F = 8.0; P < 0.001 (HI, Sn > C)

way analysis of variance with Bonferroni post-hoc analysis and with the analysis of covariance to control the effect of gender (covariate). The difference among HI, Sn and C groups (independent variable) in gender distribution was calculated using the v2 analysis. Logistic regression (stepwise method) was used to evaluate the effect of gender and temperament (indipendend variables) on the prediction of the obsessivecompulsive group (dependent variable) of each patient. Three logistic regressions were performed: in each analysis the dependent variable was C versus Sn group, C versus HI group or Sn versus HI group, respectively. Data are presented as means standard deviation. All statistical analyses were performed using SPSS v. 14.0.

j Predictors of classification as HI, Sn or C subject (Table 2)

Logistic regression analysis showed that HA, P and NS were predictors of classication as HI, Sn or C subject, whereas gender and RD did not show any effect on grouping individuals. In fact, the belonging to HI group, compared to C group, was predicted by higher levels of NS, HA and P, and, compared to Sn group, by high levels of HA and NS. Moreover, the denition of Sn subject, in comparison with C subject, was predicted by higher levels of HA and P.

Results
j Sample
The study population included 2,775 high-school students, 1,393 males (51%) and 1,382 females (49%), aged 1621 years (mean 17.4 1.0), attending the last

Discussion
Limitations. This study has some limitations. First, the use of questionnaires to assess symptomatology can be faulty unless it is followed by an interview and an objective estimation [25]. In this study, we used self-administered questionnaires alone because

Table 2 Effect of gender and temperament dimensions in predicting the belonging of a subject to High Interference (HI),Supernormal (Sn) and normal control (C) groups Obsessive-compulsive groups C (0) versus Sn (1) b Gender NS HA RD P v2 wald 2.8 3.1 17.8 2.3 12.9 P 0.09 0.07 <0.001 0.12 <0.001 OR IC 95% C (0) versus HI (1) b 0.04 0.15 0.15 v2 wald 2.1 5.3 77.4 1.1 10.2 P 0.13 0.02 <0.001 0.29 0.001 OR 1.04 1.17 1.17 IC 95% 1.001.08 1.131.21 1.061.29 Sn (0) versus HI (1) b 0.08 0.07 v2 wald 0.34 9.3 7.8 1.0 0.09 P 0.56 0.002 0.005 0.31 0.93 OR 1.09 1.07 IC 95% 1.031.15 1.021.12

0.08 0.19

1.08 1.21

1.041.13 1.091.35

NS novelty seeking; HA harm avoidance; RD reward dependence; P persistence; b coefficient of regression

C. Marchesi et al. Temperament and obsessive-compulsive symptoms in adolescents

395

informed consent to participate in the survey could be obtained only if privacy was granted to the students. The use of questionnaires involves the risk of misunderstanding questions. This possible bias was unlikely because all answers were given in the classroom, with the help of one investigator, if needed, and after detailed instructions. The use of questionnaires offers some advantages: it makes possible to screen a great number of subjects, and reduces their secretiveness (particularly in case of OCD), if the questionnaires are totally anonymous. Second, the large group of our subjects have not been screened for any other psychiatric disorder; therefore, we cannot rule out some degrees of interference on the results accounted by other psychopathological conditions (i.e. depressive disorders, anxiety disorders other than OCD, Aspergers syndrome). Third, we described the members of the Sn group as subjects with ego-syntonic obsessive symptomatology, assuming that their condition does not belong to the realm of clinical OCD but mainly to obsessive-compulsive personality disorder (OCPD) traits. This assumption has been widely questioned [3, 12, 22]; however, clinical OCD does not seems to overlap more frequently with OCPD than with other Personality Disorders [1, 19]. Ego-syntonic obsessive-compulsive symptoms have been also described as distinguishing features of more severe, nearly psychotic cases of OCD [12]. Therefore it is not clear whether the Sn group includes subjects with OCPD traits or those with a severe form of OCD or both. Given these limitations, the design of the present study does not allow any denite conclusion about the role of temperament on modulating the threshold of acceptability of OC symptoms. However some observations can be made to elicit further and more systematic investigations. 1. In this study, we found that HI subjects scored higher on HA dimension than C, with Sn group standing in a median position and differing signicatively from the other group. High HA seems to be a temperamental characteristic of people developing anxiety disorders [9], including OCD [2, 1416, 20, 24]. Using a different model of temperament, Ivarsson and Winge-Westholm [13] identied a subgroup characterized by an inhibited/shy temperament (low levels of emotionality, activity and sociability) in 83 OCD children or adolescents. This classication agrees with our data of high levels of HA in adolescents of the HI group. In our study, the levels of HA predicted which group an adolescent belonged to: in fact, increasing levels of HA predicted not only whether an individual

was classied without (C subjects) or with OC symptoms (HI or Sn subject) but also whether he/she developed ego-syntonic (Sn) or ego-dystonic (HI) symptoms. This nding may indicate a continuum in HA dimension, with increase of symptom severity until clinical manifestation of OCD above a threshold. Supporting this hypothesis, a previous study [15] showed that HA levels were positively related to symptom severity in adult OCD patients. 2. In this study, high P also was a common feature shown by ego-dystonic (HI) and ego-syntonic (Sn) subjects. Moreover, high P was a predictor of classication in groups of subjects with OC symptoms, regardless of the threshold of acceptability of symptoms. Therefore our data showed that subjects characterized by tenacity (which is attributed to people that are industrious and diligent, hard-working, ambitious and overachiever, perseverant and perfectionist) are predisposed to develop OC features (either ego-dystonic or ego-syntonic). These results represent an unpredicted nding, because none of the previous studies performed in adult patients reported similar data. This nding might be related to the peculiarity of our sample: we evaluated adolescent scholars and the discrepancy between these subjects and clinical population might be due to the effect of many variables which are very common in clinical cases of OCD (use of drugs, psychotherapy, demoralization, comorbid conditions, etc...) and may affect the temperament evaluation. Moreover, the high number of subjects included in the study may produce small but statistically signicant differences in temperament scores. However, our data conrm the description of a compulsive temperamental pattern [7] which is characterized by high levels of HA and P and low levels of NS. 3. NS scores were lower in Sn group than in HI and C groups, which showed similar values. Low levels of NS were previously found in some studies of OCD adult patients [2, 14, 15, 20], but not in others [11, 16, 24]. Low levels of NS describes individuals who are resistant to changes in familiar routine, orderly, disciplined, reserved , controlled and reective. This condition may have some protective effect against the development of interfering OC symptoms. In agreement with this hypothesis, we found that levels of NS represents the best predictor of the development of interfering or non-interfering OC symptoms. 4. Levels of RD higher than normal were found in the HI and Sn subjects. However, this difference disappeared when the gender effect was controlled, suggesting that high levels of RD were associated with female gender, which was more frequent in HI and Sn groups than in C group.

396

European Child & Adolescent Psychiatry (2008) Vol. 17, No. 6 Steinkopff Verlag 2008

Conclusions
Altogether, our data suggest that subjects characterized by pessimistic worry in anticipation of future problems, passive avoidant behaviour, rapid fatigability (high HA) and irresoluteness, ambitiousness, perseverance, perfectionism, enduring feelings of frustration (high P) might develop OC symptoms. Whether OC symptoms become ego-syntonic (Sn) or ego-dystonic (HI) seems to depend on levels of HA and NS: when low levels of NS are combined with high levels of HA, the individual may develop ego-systonic symptoms; in contrast, when the

highest levels of HA are associated with increasing levels of NS, subject may develop ego-dystonic symptoms. Therefore, our data suggest that low NS, in the presence of high HA and P, might protect people from the development of interfering OC symptoms. This might be due to prevention of exploratory and active behaviours that may elicit loss of control on symptoms. This temperamental pattern, dened by Cloninger [5, 7] as compulsive, was found to be closely related to DSM-III-R OC personality traits in adolescent [17]. Therefore, the temperamental pattern of our Sn adolescents suggests that the low NS and high HA and P predispose people to develop OCDP.

References
1. Baer L, Jenike MA, Ricciardi JN II, Holland AD, Seymour RJ, Minichiello WE, Buttolph ML (1990) Standardized assessment of personality disorders in obsessive-compulsive disorder. Arch Gen Psychiatry 47:826830 2. Bejerot S, Schlette P, Ekselius L, Adolfsson R, von Knorring L (1998) Personality disorders and relationship to personality dimensions measured by the Temperament and Character Inventory in patients with obsessivecompulsive disorder. Acta Psychiatr Scand 98:243249 3. Berg C, Rapoport JL, Flament MF (1986) The Leyton obsessional inventory-child version. J Am Acad Child Adolesc Psychiatry 25:8491 4. Cloninger CR (1986) A unied biosocial theory of personality and its role in the development of anxiety states. Psychiatr Dev 3:167226 5. Cloninger CR (1987) A systematic method for clinical description and classication of personality variance: a proposal. Arch Gen Psychiatry 44:573 588 6. Cloninger CR (1987) The tridimensional personality questionnaire. Department of Psychiatry, Washington University School of Medicine, St Louis, MO 7. Cloninger CR (1996) Assessment of the impulsive-compulsive spectrum of behaviour by the seven-factor model of temperament and character. In: Oldham JM, Hollander E, Skodol AE (eds) Impulsivity and compulsivity. American Psychiatric Association. Washington, DC 8. Cloninger CR, Martin RL, Guze SB, Clayton MD (1990) The empirical structure of psychiatric comorbidity and its theoretical signicance. In: Maser JD, Cloninger CR (eds) Comorbidity of mood and anxiety disorders. American Psychiatric Press, Washington, DC 9. Cloninger CR, Svrakic DM, Przybeck TR (1993) A psychobiological model of temperament and character. Arch Gen Psychiatry 50:975990 10. Flament MF, Whitaker A, Rapoport JL, Davies M, Berg CZ, Kalikow K, Sceery W, Schaffer D (1988) Obsessivecompulsive disorder in adolescence: an epidemiological study. J Am Acad Child Adolesc Psychiatry 27:764771 11. Gothelf D, Aharonovsky O, Horesh N, Carty T, Apter A (2004) Life events and personality factors in children and adolescents with obsessive-compulsive disorder and other anxiety disorders. Compr Psychiatry 45:192198 12. Insel TR, Akiskal HS (1986) Obsessivecompulsive disorder with psychotic features: a phenomenological analysis. Am J Psychiatry 14:15271533 13. Ivarsson T, Winge-Westholm C (2004) Temperamental factors in children and adolescents with obsessive-compulsive disorder (OCD) and in normal controls. Eur Child Adolesc Psychiatry 13:365372 14. Kusunoki K, Sato T, Taga C, Yoshida T, Komori K, Narita T, Hirano S, Iwata N, Ozaki N (2000) Low novelty-seeking differentiates obsessive-compulsive disorder from major depression. Acta Psychiatric Scand 101:403405 15. Lyoo IK, Lee DW, Kim YS, Kong SW, Kwon JS (2001) Patterns of temperament and character in subjects with obsessive-compulsive disorder. J Clin Psychiatry 62:637641 16. Lyoo IK, Yoon T, Kang DH, Kwon JS (2003) Patterns of changes in temperament and character inventory scales in subjects with obsessive-compulsive disorder following a 4-month treatment. Acta Psychiatric Scand 107:298 304 17. Maggini C, Ampollini P, Marchesi C, Gariboldi S, Cloninger CR (2000) Relationships between Tridimensional Personality Dimensions and DSM-IIIR Personality Traits in Italian Adolescents. Compr Psychiatry 41:426431 18. Maggini C, Ampollini P, Gariboldi S, Cella PL, Polizza L, Marchesi C (2001) The Parma high school epidemiological survey: obsessive-compulsive symptoms. Acta Psychiatric Scand 103:441 446 19. Mavissakalian M, Hamann MS, Jones B (1990) Correlates of DSM-III personality disorder in obsessive-compulsive disorder. Compr Psychiatry 31:481489 20. Pfohl B, Black DW, Noyes E, Kelly M, Blum N (1990) A test of the Tridimensional personality theory: association with diagnosis and platelet imipramine binding in obsessive-compulsive disorder. Biol Psychiatry 28:4146 21. Rachman SJ, de Silva P (1978) Abnormal and normal obsessions. Behav Res Therapy 16:233248 22. Rapoport JL (1986) Childhood obsessive-compulsive disorder. J Child Psychol Psychiatry 27:289295 23. Rasmussen SA, Eisen JL (1992) The epidemiology and clinical features of obsessive-compulsive disorder. Psychiatr Clin North Am 15:743757 24. Richter MA, Summerfeldt LJ, Joffe RT, Swinson RP (1996) The Tridimensional Personality Questionnaire in obsessivecompulsive disorder. Psychiatr Res 65:185188 25. Thomsen PH (1993) Obsessive-compulsive disorder in children and adolescents. Self-reported obsessivecompulsive behaviour in pupils in Denmark. Acta Psychiatr Scand 88:212217