- Email: [email protected]
Temperament in bipolar illness: impact on prognosis
Journal of Affective Disorders 56 (1999) 103–108 www.elsevier.com / locate / jad
Research report
Temperament in bipolar illness: impact on prognosis ´ ˆ ´ ` Verdoux a , Chantal Henry a , *, Jerome Lacoste a , Frank Bellivier b , Helene a b ,c Marc L. Bourgeois , Marion Leboyer a
b
´ , 33076 Bordeaux Cedex, France Service Universitaire de Psychiatrie, CHS Charles Perrens, 121 rue de la Bechade ´ et les Conduites Adaptatives, CNRS UMR 7593, Hopital ˆ ´ ´ ` , Paris, Laboratoire de Recherche sur les Personnalites Pitie-Salpetriere France c ˆ ˆ ´ ´ ` , Paris, France de Paris, Hopital Pitie-Salpetriere Service de Psychiatrie Adulte, Assistance Publique–Hopitaux Received 21 July 1998; received in revised form 25 October 1998; accepted 20 November 1998
Abstract Objective: The present study was designed to investigate the relations between temperament and outcome in bipolar illness. Methods: Seventy-two patients presenting with bipolar type I disorder were recruited from consecutive admissions and evaluated when euthymic. The criteria developed by Akiskal and Mallya (Criteria for the ‘soft’ bipolar spectrum: treatment implications. Psychopharmacol. Bull. 1987;23:68–73) were used to assess both depressive (DT) and hyperthymic temperaments (HT) in a dimensional approach. Results: Multiple regression analysis showed that a higher DT score or a lower HT score were significantly associated with a greater number of episodes. Furthermore, a higher DT score was strongly associated with a higher percentage of major depressive episodes. Conversely, a higher HT score was associated with a trend to manic rather than depressive episodes. Suicide attempts appeared more frequent in the history of patients presenting with higher DT scores. Conclusions: Our findings strengthen the hypothesis that temperament is one of the main variables accounting for some features in the clinical evolution of bipolar disorder such as polarity of episodes. Furthermore, these findings are consistent with the hypothesis of a trait-state continuum between personality and affective episodes. 1999 Elsevier Science B.V. All rights reserved. Keywords: Bipolar disorder; Prognosis; Hyperthymic temperament; Depressive temperament; Suicide attempts; Manic episodes; Depressive episodes
1. Introduction Kraepelin (1921) was among the first author to pay systematic attention to the premorbid characteristics of patients suffering from affective disor*Corresponding author. Tel.: 1 33-556-563448; fax: 1 33-556563546. E-mail address: [email protected] (C. Henry)
ders. In his attempt to account for the recurrence of affective episodes, he postulated the existence of enduring personality characteristics, from which the affective states arose. Thus, depressive, manic, irritable and cyclothymic personalities were considered the temperamental bases of the full-blown forms of the illness. The Kraepelin (1921) hypothesis was based on the observation that affective temperaments occurred in the premorbid histories of most of
0165-0327 / 99 / $ – see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S0165-0327( 98 )00219-5
104
C. Henry et al. / Journal of Affective Disorders 56 (1999) 103 – 108
manic-depressive probands who returned to their basic temperament rather than to ‘normality’ on remission; furthermore, such temperaments, without progression to full-blown illness, were over-represented in the biologic relatives of manic-depressive probands. The work of Schneider (1958) extended Kraepelin’s observations and led to descriptions of the depressive and hypomanic types which are the two basic temperaments in the Kraepelinian scheme. Kretschmer (1936) postulated a continuum between the cyclothymic character and manic-depressive psychosis. More recently, some authors (Von Zerssen and Possl, 1990; Akiskal and Akiskal, 1992) suggested that one of the major point of interest with temperament is to be able to explore the clinical and predictive significance of subaffective traits in order to understand the premorbid traits and interepisodic manifestations of recurrent mood disorders within a spectrum of bipolar illness. Thus, Akiskal (1995) showed that dysthymic, cyclothymic, and hyperthymic temperaments represent putative developmental pathways to bipolarity in childhood and adolescence with clinically ascertained depressions. Other studies have focused on the fact that temperaments may broaden the spectrum of bipolar disorders. Cassano et al. (1989) showed that in unipolar disorder with hyperthymic temperament, family history and sex distribution have a closer resemblance to those of bipolar I and II disorders rather than pure unipolar disorder (without hyperthymic temperament). Therefore, the bipolar spectrum for these authors includes hyperthymic or cyclothymic temperaments associated with major depressions (Akiskal, 1983; Akiskal et al., 1983; Cassano et al., 1992). However, only few studies have investigated the relationships between hyperthymic or depressive temperaments and the main characteristics of bipolar disorder. The aim of the present study is to explore the impact of temperamental status differences on the evolution of bipolar illness.
2. Methods
2.1. Subjects Seventy-two patients presenting with bipolar I
disorder were recruited from consecutive admissions to a unit serving an urban geographic area in the Bordeaux psychiatric hospital. All patients gave written informed consent. Patients were interviewed by trained psychiatrists (CH and JL) with a French version of the Diagnosis Interview for Genetic Studies (Nurnberger et al., 1994) providing DSM IV Axis I diagnoses. Best estimate based on consensus between the interviewer and an independent psychiatrist was obtained for the following anamnestic data: number of previous episodes, number of manic and depressive episodes, type of first episode, age at first hospitalization, history of suicide attempts, and sociodemographic data. At the time of interview, which usually took place at the end of the hospital stay, patients had recovered from their most recent episode.
2.2. Rating scales To evaluate temperamental aspects, we used depressive and hyperthymic temperament criteria developed by Akiskal and Mallya (1987). We used the French version of the Semi-structured Interview of these temperaments based upon the University of Tennessee modification (Akiskal and Mallya, 1987). This French version has criteria regarding the subaffective feature of temperaments like sleep duration and energy level (Bourgeois et al., 1997). These two interviews (one for hyperthymic and one for depressive temperament) were composed of 22 items about usual mood, cognition, psychomotricity, personal interrelations, attitudes to social norms, sleep needs, and sexual appetite. Answers were selected by the patient himself and not by the scorer. One point was counted per positive item. All patients were assessed using the two interviews and had a score for both temperaments. These tools are usually used to distinguish sub-groups of patients. In this case, each temperament requires the presence of at least five items and both diagnoses are mutually exclusive. Some patients cannot be classified in any of category because they lack the number of items. In a study by Perugi et al. (1990) only 59% of bipolar type I patients could be characterized on their baseline temperament with this approach. We chose to consider this temperamental characteristic as a dimension rather than a category. Indeed, the analysis of score distribution in both temperament scales does
C. Henry et al. / Journal of Affective Disorders 56 (1999) 103 – 108
not evidence a bi- or tri-modal distribution allowing us to establish non-arbitrary thresholds in the constitution of categories. Furthermore, the advantage of a dimensional approach is that no patient is excluded. Only these two temperament subtypes were used because they represent two opposing symmetrical traits.
2.3. Statistical analysis Statistical analysis were carried out using SPSS software (Norusis, 1992). Associations between categorical variables and temperaments dimensions scores were examined using logistic regression, yielding odds ratios (OR) and 95% confidence intervals (CI). Multiple regression analyses yielding regression coefficients (B) and 95% confidence intervals (CI) were used to examine the relationships between continuous variables and dimensions scores. Distributions of continuous variables were examined, and log-transformations or square-transformations were made in order to remove skewness where appropriate. Regression models were subsequently computed adjusted for potential confounding variables, which were selected a priori and not on the basis of statistical significant association with the dependent and explanatory variables (Clayton and Hills, 1993). Residuals were examined for evidence of non-linearity. Departure from linearity was also examined by adding square clinical rating score terms to the model and assessing subsequent improvement in the model fit.
3. Results
3.1. Sample characteristics The sample was composed of 27 (37.5%) males and 45 (62.5%) females, with a mean age of 41.7 years (S.D. 13.9; range 19–77). Educational level was less than ‘baccalaureate’ ( , 12 years of education) for 30 (41.7%) patients and equal to or more than ‘baccalaureate’ ( $ 12 years of education) for 42 (58.3%) patients. The occupational status was ‘unemployed’ or ‘disability pension’ for 26 (36.1%) patients, and ‘currently working’, ‘student’ or ‘retired’ for 46 (63.9%) patients. Concerning marital status, 33 (45.8%)
105
patients had always been single, and 39 (54.2%) patients were married, divorced, or widowed. By defining the onset of the first symptoms meeting DSM IV criteria for an affective episode, we found that the mean age for the first depressive episode in this population was 26.9 (S.D. 9) and 30 (S.D. 10.8) for the first manic episode. The proportion of bipolar patients who began their illness with a manic episode was 31.9%, and was 58.3% for those whose first episode was depressive (no reliable information was available for 9.8% patients). Patients were clinically euthymic when included as confirmed by the mean MADRS scores 5 2.3 (S.D. 2.9) (Montgomery and Asberg, 1979) and the mean Bech and Rafaelsen Mania Rating Scale scores 5 2.7 (S.D. 2.2) (Bech et al., 1978).
3.2. Associations between characteristics of illness and temperaments As indicated in Table 1, logistic or multiple regressions indicated that hyperthymic temperament (HT) scores, as well as depressive temperament (DT) scores were significantly correlated with some clinical characteristics of bipolar disorders. The total number of manic or depressive episodes was positively correlated to the DT score (B 5 0.06; CI (0.02; 0.11); P 5 0.01) and negatively to the HT score (B 5 2 0.05; CI (20.09; 20.009); P 5 0.02). This trend was not abolished after adjustment for age, sex and educational level. Furthermore, the percentage of manic episodes was positively correlated to the HT score (B 5 0.008; CI (0.00; 0.02); P 5 0.04), whereas the negative correlation with DT score was highly significant (B 5 2 0.02; CI (20.03; 2 0.01); P 5 0.0005) even after adjustment for sex, age and educational level. A depressive temperament could therefore be considered predictive for an increased frequency of episodes, and for a depressive rather than manic polarity of these episodes. Conversely, elevated HT scores was associated with a strong trend to have manic rather than depressive episodes. Neither the type of the first episode, nor the age at first hospitalization were associated with HT or DT temperament scales. In contrast, the history of suicide attempts was influenced by temperament status since there was a significant association between depressive scores and suicide attempts (OR 5
106
C. Henry et al. / Journal of Affective Disorders 56 (1999) 103 – 108
Table 1 Association between hyperthymic or depressive temperaments and clinical features of the manic-depressive illness: logistic or multiple regressions. Association with Total number of episodes Hyperthymic score Depression score Percentage manic episodes Hyperthymic score Depression score Type of first episode Hyperthymic score Depression score Age at first hospitalization Hyperthymic score Depression score History of suicide attempt Hyperthymic score Depression score Occupational status b Hyperthymic score Depression score Marital status c Hyperthymic score Depression score
Unadjusted
Adjusted for age, sex, educational level a
B 5 2 0.05 B 5 0.06
p 5 0.02 p 5 0.01
B 5 2 0.03 B 5 0.04
p 5 0.07 p 5 0.06
B 5 0.008 B 5 2 0.02
p 5 0.04 p 5 0.0005
B 5 0.007 B 5 2 0.01
p 5 0.09 p 5 0.002
OR 5 1.01 OR 5 0.97
p 5 0.70 p 5 0.62
OR 5 1.04 OR 5 0.94
p 5 0.46 p 5 0.38
B 5 2 0.003 B 5 0.007
p 5 0.68 p 5 0.43
B 5 0.002 B 5 0.0
p 5 0.82 p 5 0.92
OR 5 0.88 OR 5 1.26
p 5 0.02 p 5 0.002
OR 5 0.88 OR 5 1.16
p 5 0.04 p 5 0.003
OR 5 0.97 OR 5 0.96
p 5 0.62 p 5 0.61
OR 5 0.99 OR 5 0.94
p 5 0.84 p 5 0.34
OR 5 0.92 OR 5 1.08
p 5 0.10 p 5 0.18
OR 5 0.95 OR 5 1.01
p 5 0.35 p 5 0.87
a
, vs $ 12 years of education. The associations with occupational level were not adjusted for this variable. Unemployed / disability pension vs currently working, student or retired. c Always single vs other (married, divorced, widowed). b
1.26; CI (1.09; 1.46); P 5 0.002). This association remained significant after adjustment for age, sex and educational level, and also after further adjustment for the percentage of depressive episodes. Finally, no association was found between HT or DT scores and occupational or marital status.
4. Discussion In this study of 72 patients with bipolar type I disorder, we show that a high score on the DT score is associated with a greater lifetime number of episodes, a high percentage of major depressive episodes and a high history of suicide attempts. Conversely, a high HT score is associated with a strong trend to have manic rather than depressive episodes. Thus, our study shows a continuum between temperament and polarity of episodes and a better prognosis for patients presenting with more hyperthymic traits.
We have a priori chosen to consider temperaments as dimensions rather than categories in order to avoid arbitrary thresholds. Since this approach has not been previously used, our results have to be cautiously compared to those of previous studies. Assessment of temperaments was carried out when patients had recovered, as confirmed by low scores on the MADRS and Bech mania scales. It is therefore unlikely that the present findings may be explained by state-dependent variables. In addition, Klein (1990) showed longitudinal stability and a lack of thymic state-dependency with the depressive temperament Our results are consistent with data previously reported. Perugi et al. (1990) investigated possible gender-mediated clinical expressions of depression, focusing on temperamental characteristics and the longitudinal aspects of mood disorder. The incidence of depressive episodes was higher in women, as was that of depressive temperament. By contrast, males had higher rates of hyperthymic temperament and
C. Henry et al. / Journal of Affective Disorders 56 (1999) 103 – 108
hypomanic episodes. The sample for this study included a low percentage of Bipolar I disorder, only 5.4%, and the majority were patients with single or recurrent depressives episodes. For the authors, temperamental differences might account for some of the observed gender differences in the rates of depression and its subtypes. Some studies have shown that temperaments influence the incidence of mixed episodes. Indeed, Dell’Osso et al. (1991) showed that mixed states are more likely to arise from a depressive temperament. This supports the Akiskal (1992) clinical observation that mixed states tend to arise when temperament and episodes are of opposing polarity. Dell’Osso et al. (1991) concluded that the mixed state tends to have a distinct longitudinal pattern of manic-depressive illness. Instead of postulating the existence of such a sub-category, we could also assume that this pattern is underlied by the depressive temperament which may influence the evolution and the clinical picture. On the other hand, these temperamental differences could explain why most studies on bipolar patients’ personality do not find differences when comparing with control subjects (Hirschfeld and Klerman, 1979; Liebowitz et al., 1979; Goodwin and Jamison, 1990). Most of these studies compare introverted versus extroverted traits which are respectively very close to depressive and hyperthymic temperaments. Given that bipolar patients represent a very heterogeneous group for drive, affect, and emotions, all of which define their temperament, and that these temperaments could be opposed, then the negative results of these studies comparing median values may be understood. The parallel association between temperament and polarity of episodes raises the question whether these temperamental differences are cause or consequence of the episodes. Definitive data on this question are presently unavailable. However, much evidence supports the hypothesis that the temperamental attributes develop early in life and represent long-term functioning of the individual (Musetti et al., 1989). Studies in both patients and normal populations (Depue et al., 1981; Akiskal, 1984) have shown that these traits often exist in the absence of major affective episodes. In another 3–4 year prospective study on adolescent and pre-pubertal offspring and
107
siblings of adults with bipolar affective disorder, hyperthymic and depressive temperaments were observed before any superimposed episodes occurred (Akiskal et al., 1985). Strelau (1983) defined temperament as ‘the relatively stable features of the organism, primarily biologically determined, as revealed in the formal traits of reactions which form the energy level and temporal characteristics of behavior’. Klein et al. (1988) demonstrated dysthymia among the offspring of probands with major affective illness. Klein et al. (1985) also demonstrated the occurrence of cyclothymia among the adolescent offspring of bipolar adults. Furthermore, among monozygotic twins of patients with bipolar disorder who are not themselves affected, over half suffer from temperamental disturbances (Bertelsen et al., 1977). Thus, it could be assumed that temperamental attributes would represent predisposing familial-risk factors for affective illness. Further investigations regarding comorbidity, syndromal phenomenology, treatment response and temperament in relatives are now required to strengthen the hypothesis of a trait-state continuum between personality and affective episodes, which could both represent the expression of the same genetic or constitutional endowment.
Acknowledgements ˆ This work was supported by a grant from RhonePoulenc Rorer (JL) and Assistance Publique.
References Akiskal, H.S., 1983. The bipolar spectrum: New concepts in classification and diagnosis. In: L. Grinspoon (Ed.), Psychiatry Update: The American Psychiatric Association Annual Review. American Psychiatric Press, Washington, DC, pp. 271–292. Akiskal, H.S., 1984. Characterological manifestations of affective disorders. Toward a new conceptualization. Integr. Psychiatry 2, 83–88. Akiskal, H.S., 1992. The distinctive mixed states of bipolar I, II, and III. Clin. Neuropharmacol. 15, 632A. Akiskal, H.S., 1995. Developmental pathways to bipolarity: are juvenile-onset depressions pre-bipolar? J. Am. Acad. Child Adolesc. Psychiatry 34, 754–763.
108
C. Henry et al. / Journal of Affective Disorders 56 (1999) 103 – 108
Akiskal, H.S., Mallya, G., 1987. Criteria for the ‘soft’ bipolar spectrum: treatment implications. Psychopharmacol. Bull. 23, 68–73. Akiskal, H.S., Akiskal, K., 1992. Cyclothymic, hyperthymic and depressive temperaments as subaffective variants of mood disorders. In: Tasman, A. (Ed.), American Psychiatric Association Review. American Psychiatric Press, Washington, DC, pp. 43–62. Akiskal, H.S., Hirschfeld, R.M.A., Yerevanian, B.I., 1983. The relationship of personality to affective disorders: a critical review. Arch. Gen. Psychiatry 40, 801–810. Akiskal, H.S., Downs, J., Jordan, P., et al., 1985. Affective disorders in the referred children and younger sibling of manic depressives: mode of onset and prospective course. Arch. Gen. Psychiatry 42, 996–1007. Bech, P., Rafaelsen, O.J., Kramp, P., Bolwig, T.G., 1978. The Mania Rating Scale. Scale construction and inter-observer agreement. Neuropharmacology 17, 430–441. Bertelsen, A., Harvald, D., Hauge, M., 1977. A Danish twin study of manic depressive disorders. Br. J. Psychiatry 130, 330–351. Bourgeois, M.L., Hantouche, E.G., Akiskal, H.S., 1997. The EPIMAN and EPIDEP French studies of bipolarity. Preliminary results. J. Bipolar Disord. 1, 13–19. Cassano, G.B., Akiskal, H.S., Musetti, L., Perugi, G., Soriani, A., Mignani, V., 1989. Psychopathology, temperament, and past course in primary major depressions. Psychopathology 22, 278–288. Cassano, G.B., Akiskal, H.S., Perugi, G., Musetti, L., Savino, M., 1992. The importance of measures of affective temperaments in genetic studies of mood disorders. J. Psychiat. Res. 26, 257–268. Clayton, D., Hills, M., 1993. Choise and interpretation of models. In: Statistical models in epidemiology. Oxford University Press, Oxford, pp. 271–281. Dell’Osso, L., Placidi, G.F., Nassi, R., Freer, P., Cassano, G.B., Akiskal, H.S., 1991. The manic-depressive mixed state: familial, temperamental and psychopathologic characteristics in 108 female inpatients. Eur. Arch. Psychiatry Clin. Neurosci. 240, 234–239. Depue, R., Slater, J.F., Wolfstetter-Kausch, H., 1981. A behavioral paradigm for identifying persons at risk for bipolar depressive disorders: a conceptual framework and five validating studies. J. Abnorm. Psychol. 90, 381–437. Goodwin, F.K., Jamison, K.R., 1990. Manic-Depressive Illness. Oxford University Press, New York.
Hirschfeld, R.M.A., Klerman, G.L., 1979. Personality attributes and affective disorders. Am. J. Psychiatry 136, 67–70. Klein, D.N., Depue, R.A., Slater, J.F., 1985. Cyclothymia in the adolescent offspring of parents with bipolar affective disorder. J. Abnorm. Psychol. 94, 115–127. Klein, D.N., 1990. Depressive personality: reliability, validity, and relation to dysthymia. J. Abnorm. Psychol. 99, 412–421. Klein, D.N., Taylor, E.B., Harding, K., et al., 1988. Double depression and episodic major depression: demographic, clinical, familial, personality, and socioenvironmental characteristics and short-term outcome. Am. J. Psychiatry 145, 1226– 1231. Kraepelin, E., 1921. Manic-Depressive Insanity and Paranoia. Livingstone, Edinburgh. Kretschmer, E., 1936. Physique and Character (Miller, E., trans.). Kegan Paul (Eds.), London. Liebowitz, M.R., Stallone, F., Dunner, D.L., Fieve, R.R., 1979. Personality features of patients with primary affective disorder. Acta Psychiatr. Scand. 60, 214–224. Montgomery, S.A., Asberg, M., 1979. A new depressed scale designed to be sensitive to change. Br. J. Psychiatry 134, 382–389. Musetti, L., Perugi, G., Soriani, A., et al., 1989. Depression before and after age 65. A re-examination. Br. J. Psychiatry 155, 330–336. Norusis, M.J., 1992. SPPS Inc. SPSS for Windows 5.0.1. SPSS Inc., Chicago. Nurnberger, J.I., Blehar, M.C., Kaufman, C.A., York-Cooler, C., Simpson, S.G., Harkavy-Freedman, J., Severe, J.B., Malaspina, D., Reich, T., 1994. Diagnostic intreview for genetic studies. Arch. Gen. Psychiatry 51, 849–864. Perugi, G., Musetti, L., Simonini, E., Piagentini, F., Cassano, G.B., Akiskal, H.S., 1990. Gender-mediated clinical features of depressive illness: the importance of temperamental differences. Br. J. Psychiatry 157, 835–841. Schneider, K., 1958. Psychopathic Personalities. (Hamilton, M.W., trans.). Cassel, London. Strelau, J., 1983. Temperament, Personality and Arousal. Academic Press, London. Von Zerssen, D., Possl, J., 1990. The premorbid personality of patients with different subtypes of an affective illness. Statistical analysis of blind assignment of case history data to clinical diagnosis. J. Affect. Disord. 18, 39–50.
Research report
Temperament in bipolar illness: impact on prognosis ´ ˆ ´ ` Verdoux a , Chantal Henry a , *, Jerome Lacoste a , Frank Bellivier b , Helene a b ,c Marc L. Bourgeois , Marion Leboyer a
b
´ , 33076 Bordeaux Cedex, France Service Universitaire de Psychiatrie, CHS Charles Perrens, 121 rue de la Bechade ´ et les Conduites Adaptatives, CNRS UMR 7593, Hopital ˆ ´ ´ ` , Paris, Laboratoire de Recherche sur les Personnalites Pitie-Salpetriere France c ˆ ˆ ´ ´ ` , Paris, France de Paris, Hopital Pitie-Salpetriere Service de Psychiatrie Adulte, Assistance Publique–Hopitaux Received 21 July 1998; received in revised form 25 October 1998; accepted 20 November 1998
Abstract Objective: The present study was designed to investigate the relations between temperament and outcome in bipolar illness. Methods: Seventy-two patients presenting with bipolar type I disorder were recruited from consecutive admissions and evaluated when euthymic. The criteria developed by Akiskal and Mallya (Criteria for the ‘soft’ bipolar spectrum: treatment implications. Psychopharmacol. Bull. 1987;23:68–73) were used to assess both depressive (DT) and hyperthymic temperaments (HT) in a dimensional approach. Results: Multiple regression analysis showed that a higher DT score or a lower HT score were significantly associated with a greater number of episodes. Furthermore, a higher DT score was strongly associated with a higher percentage of major depressive episodes. Conversely, a higher HT score was associated with a trend to manic rather than depressive episodes. Suicide attempts appeared more frequent in the history of patients presenting with higher DT scores. Conclusions: Our findings strengthen the hypothesis that temperament is one of the main variables accounting for some features in the clinical evolution of bipolar disorder such as polarity of episodes. Furthermore, these findings are consistent with the hypothesis of a trait-state continuum between personality and affective episodes. 1999 Elsevier Science B.V. All rights reserved. Keywords: Bipolar disorder; Prognosis; Hyperthymic temperament; Depressive temperament; Suicide attempts; Manic episodes; Depressive episodes
1. Introduction Kraepelin (1921) was among the first author to pay systematic attention to the premorbid characteristics of patients suffering from affective disor*Corresponding author. Tel.: 1 33-556-563448; fax: 1 33-556563546. E-mail address: [email protected] (C. Henry)
ders. In his attempt to account for the recurrence of affective episodes, he postulated the existence of enduring personality characteristics, from which the affective states arose. Thus, depressive, manic, irritable and cyclothymic personalities were considered the temperamental bases of the full-blown forms of the illness. The Kraepelin (1921) hypothesis was based on the observation that affective temperaments occurred in the premorbid histories of most of
0165-0327 / 99 / $ – see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S0165-0327( 98 )00219-5
104
C. Henry et al. / Journal of Affective Disorders 56 (1999) 103 – 108
manic-depressive probands who returned to their basic temperament rather than to ‘normality’ on remission; furthermore, such temperaments, without progression to full-blown illness, were over-represented in the biologic relatives of manic-depressive probands. The work of Schneider (1958) extended Kraepelin’s observations and led to descriptions of the depressive and hypomanic types which are the two basic temperaments in the Kraepelinian scheme. Kretschmer (1936) postulated a continuum between the cyclothymic character and manic-depressive psychosis. More recently, some authors (Von Zerssen and Possl, 1990; Akiskal and Akiskal, 1992) suggested that one of the major point of interest with temperament is to be able to explore the clinical and predictive significance of subaffective traits in order to understand the premorbid traits and interepisodic manifestations of recurrent mood disorders within a spectrum of bipolar illness. Thus, Akiskal (1995) showed that dysthymic, cyclothymic, and hyperthymic temperaments represent putative developmental pathways to bipolarity in childhood and adolescence with clinically ascertained depressions. Other studies have focused on the fact that temperaments may broaden the spectrum of bipolar disorders. Cassano et al. (1989) showed that in unipolar disorder with hyperthymic temperament, family history and sex distribution have a closer resemblance to those of bipolar I and II disorders rather than pure unipolar disorder (without hyperthymic temperament). Therefore, the bipolar spectrum for these authors includes hyperthymic or cyclothymic temperaments associated with major depressions (Akiskal, 1983; Akiskal et al., 1983; Cassano et al., 1992). However, only few studies have investigated the relationships between hyperthymic or depressive temperaments and the main characteristics of bipolar disorder. The aim of the present study is to explore the impact of temperamental status differences on the evolution of bipolar illness.
2. Methods
2.1. Subjects Seventy-two patients presenting with bipolar I
disorder were recruited from consecutive admissions to a unit serving an urban geographic area in the Bordeaux psychiatric hospital. All patients gave written informed consent. Patients were interviewed by trained psychiatrists (CH and JL) with a French version of the Diagnosis Interview for Genetic Studies (Nurnberger et al., 1994) providing DSM IV Axis I diagnoses. Best estimate based on consensus between the interviewer and an independent psychiatrist was obtained for the following anamnestic data: number of previous episodes, number of manic and depressive episodes, type of first episode, age at first hospitalization, history of suicide attempts, and sociodemographic data. At the time of interview, which usually took place at the end of the hospital stay, patients had recovered from their most recent episode.
2.2. Rating scales To evaluate temperamental aspects, we used depressive and hyperthymic temperament criteria developed by Akiskal and Mallya (1987). We used the French version of the Semi-structured Interview of these temperaments based upon the University of Tennessee modification (Akiskal and Mallya, 1987). This French version has criteria regarding the subaffective feature of temperaments like sleep duration and energy level (Bourgeois et al., 1997). These two interviews (one for hyperthymic and one for depressive temperament) were composed of 22 items about usual mood, cognition, psychomotricity, personal interrelations, attitudes to social norms, sleep needs, and sexual appetite. Answers were selected by the patient himself and not by the scorer. One point was counted per positive item. All patients were assessed using the two interviews and had a score for both temperaments. These tools are usually used to distinguish sub-groups of patients. In this case, each temperament requires the presence of at least five items and both diagnoses are mutually exclusive. Some patients cannot be classified in any of category because they lack the number of items. In a study by Perugi et al. (1990) only 59% of bipolar type I patients could be characterized on their baseline temperament with this approach. We chose to consider this temperamental characteristic as a dimension rather than a category. Indeed, the analysis of score distribution in both temperament scales does
C. Henry et al. / Journal of Affective Disorders 56 (1999) 103 – 108
not evidence a bi- or tri-modal distribution allowing us to establish non-arbitrary thresholds in the constitution of categories. Furthermore, the advantage of a dimensional approach is that no patient is excluded. Only these two temperament subtypes were used because they represent two opposing symmetrical traits.
2.3. Statistical analysis Statistical analysis were carried out using SPSS software (Norusis, 1992). Associations between categorical variables and temperaments dimensions scores were examined using logistic regression, yielding odds ratios (OR) and 95% confidence intervals (CI). Multiple regression analyses yielding regression coefficients (B) and 95% confidence intervals (CI) were used to examine the relationships between continuous variables and dimensions scores. Distributions of continuous variables were examined, and log-transformations or square-transformations were made in order to remove skewness where appropriate. Regression models were subsequently computed adjusted for potential confounding variables, which were selected a priori and not on the basis of statistical significant association with the dependent and explanatory variables (Clayton and Hills, 1993). Residuals were examined for evidence of non-linearity. Departure from linearity was also examined by adding square clinical rating score terms to the model and assessing subsequent improvement in the model fit.
3. Results
3.1. Sample characteristics The sample was composed of 27 (37.5%) males and 45 (62.5%) females, with a mean age of 41.7 years (S.D. 13.9; range 19–77). Educational level was less than ‘baccalaureate’ ( , 12 years of education) for 30 (41.7%) patients and equal to or more than ‘baccalaureate’ ( $ 12 years of education) for 42 (58.3%) patients. The occupational status was ‘unemployed’ or ‘disability pension’ for 26 (36.1%) patients, and ‘currently working’, ‘student’ or ‘retired’ for 46 (63.9%) patients. Concerning marital status, 33 (45.8%)
105
patients had always been single, and 39 (54.2%) patients were married, divorced, or widowed. By defining the onset of the first symptoms meeting DSM IV criteria for an affective episode, we found that the mean age for the first depressive episode in this population was 26.9 (S.D. 9) and 30 (S.D. 10.8) for the first manic episode. The proportion of bipolar patients who began their illness with a manic episode was 31.9%, and was 58.3% for those whose first episode was depressive (no reliable information was available for 9.8% patients). Patients were clinically euthymic when included as confirmed by the mean MADRS scores 5 2.3 (S.D. 2.9) (Montgomery and Asberg, 1979) and the mean Bech and Rafaelsen Mania Rating Scale scores 5 2.7 (S.D. 2.2) (Bech et al., 1978).
3.2. Associations between characteristics of illness and temperaments As indicated in Table 1, logistic or multiple regressions indicated that hyperthymic temperament (HT) scores, as well as depressive temperament (DT) scores were significantly correlated with some clinical characteristics of bipolar disorders. The total number of manic or depressive episodes was positively correlated to the DT score (B 5 0.06; CI (0.02; 0.11); P 5 0.01) and negatively to the HT score (B 5 2 0.05; CI (20.09; 20.009); P 5 0.02). This trend was not abolished after adjustment for age, sex and educational level. Furthermore, the percentage of manic episodes was positively correlated to the HT score (B 5 0.008; CI (0.00; 0.02); P 5 0.04), whereas the negative correlation with DT score was highly significant (B 5 2 0.02; CI (20.03; 2 0.01); P 5 0.0005) even after adjustment for sex, age and educational level. A depressive temperament could therefore be considered predictive for an increased frequency of episodes, and for a depressive rather than manic polarity of these episodes. Conversely, elevated HT scores was associated with a strong trend to have manic rather than depressive episodes. Neither the type of the first episode, nor the age at first hospitalization were associated with HT or DT temperament scales. In contrast, the history of suicide attempts was influenced by temperament status since there was a significant association between depressive scores and suicide attempts (OR 5
106
C. Henry et al. / Journal of Affective Disorders 56 (1999) 103 – 108
Table 1 Association between hyperthymic or depressive temperaments and clinical features of the manic-depressive illness: logistic or multiple regressions. Association with Total number of episodes Hyperthymic score Depression score Percentage manic episodes Hyperthymic score Depression score Type of first episode Hyperthymic score Depression score Age at first hospitalization Hyperthymic score Depression score History of suicide attempt Hyperthymic score Depression score Occupational status b Hyperthymic score Depression score Marital status c Hyperthymic score Depression score
Unadjusted
Adjusted for age, sex, educational level a
B 5 2 0.05 B 5 0.06
p 5 0.02 p 5 0.01
B 5 2 0.03 B 5 0.04
p 5 0.07 p 5 0.06
B 5 0.008 B 5 2 0.02
p 5 0.04 p 5 0.0005
B 5 0.007 B 5 2 0.01
p 5 0.09 p 5 0.002
OR 5 1.01 OR 5 0.97
p 5 0.70 p 5 0.62
OR 5 1.04 OR 5 0.94
p 5 0.46 p 5 0.38
B 5 2 0.003 B 5 0.007
p 5 0.68 p 5 0.43
B 5 0.002 B 5 0.0
p 5 0.82 p 5 0.92
OR 5 0.88 OR 5 1.26
p 5 0.02 p 5 0.002
OR 5 0.88 OR 5 1.16
p 5 0.04 p 5 0.003
OR 5 0.97 OR 5 0.96
p 5 0.62 p 5 0.61
OR 5 0.99 OR 5 0.94
p 5 0.84 p 5 0.34
OR 5 0.92 OR 5 1.08
p 5 0.10 p 5 0.18
OR 5 0.95 OR 5 1.01
p 5 0.35 p 5 0.87
a
, vs $ 12 years of education. The associations with occupational level were not adjusted for this variable. Unemployed / disability pension vs currently working, student or retired. c Always single vs other (married, divorced, widowed). b
1.26; CI (1.09; 1.46); P 5 0.002). This association remained significant after adjustment for age, sex and educational level, and also after further adjustment for the percentage of depressive episodes. Finally, no association was found between HT or DT scores and occupational or marital status.
4. Discussion In this study of 72 patients with bipolar type I disorder, we show that a high score on the DT score is associated with a greater lifetime number of episodes, a high percentage of major depressive episodes and a high history of suicide attempts. Conversely, a high HT score is associated with a strong trend to have manic rather than depressive episodes. Thus, our study shows a continuum between temperament and polarity of episodes and a better prognosis for patients presenting with more hyperthymic traits.
We have a priori chosen to consider temperaments as dimensions rather than categories in order to avoid arbitrary thresholds. Since this approach has not been previously used, our results have to be cautiously compared to those of previous studies. Assessment of temperaments was carried out when patients had recovered, as confirmed by low scores on the MADRS and Bech mania scales. It is therefore unlikely that the present findings may be explained by state-dependent variables. In addition, Klein (1990) showed longitudinal stability and a lack of thymic state-dependency with the depressive temperament Our results are consistent with data previously reported. Perugi et al. (1990) investigated possible gender-mediated clinical expressions of depression, focusing on temperamental characteristics and the longitudinal aspects of mood disorder. The incidence of depressive episodes was higher in women, as was that of depressive temperament. By contrast, males had higher rates of hyperthymic temperament and
C. Henry et al. / Journal of Affective Disorders 56 (1999) 103 – 108
hypomanic episodes. The sample for this study included a low percentage of Bipolar I disorder, only 5.4%, and the majority were patients with single or recurrent depressives episodes. For the authors, temperamental differences might account for some of the observed gender differences in the rates of depression and its subtypes. Some studies have shown that temperaments influence the incidence of mixed episodes. Indeed, Dell’Osso et al. (1991) showed that mixed states are more likely to arise from a depressive temperament. This supports the Akiskal (1992) clinical observation that mixed states tend to arise when temperament and episodes are of opposing polarity. Dell’Osso et al. (1991) concluded that the mixed state tends to have a distinct longitudinal pattern of manic-depressive illness. Instead of postulating the existence of such a sub-category, we could also assume that this pattern is underlied by the depressive temperament which may influence the evolution and the clinical picture. On the other hand, these temperamental differences could explain why most studies on bipolar patients’ personality do not find differences when comparing with control subjects (Hirschfeld and Klerman, 1979; Liebowitz et al., 1979; Goodwin and Jamison, 1990). Most of these studies compare introverted versus extroverted traits which are respectively very close to depressive and hyperthymic temperaments. Given that bipolar patients represent a very heterogeneous group for drive, affect, and emotions, all of which define their temperament, and that these temperaments could be opposed, then the negative results of these studies comparing median values may be understood. The parallel association between temperament and polarity of episodes raises the question whether these temperamental differences are cause or consequence of the episodes. Definitive data on this question are presently unavailable. However, much evidence supports the hypothesis that the temperamental attributes develop early in life and represent long-term functioning of the individual (Musetti et al., 1989). Studies in both patients and normal populations (Depue et al., 1981; Akiskal, 1984) have shown that these traits often exist in the absence of major affective episodes. In another 3–4 year prospective study on adolescent and pre-pubertal offspring and
107
siblings of adults with bipolar affective disorder, hyperthymic and depressive temperaments were observed before any superimposed episodes occurred (Akiskal et al., 1985). Strelau (1983) defined temperament as ‘the relatively stable features of the organism, primarily biologically determined, as revealed in the formal traits of reactions which form the energy level and temporal characteristics of behavior’. Klein et al. (1988) demonstrated dysthymia among the offspring of probands with major affective illness. Klein et al. (1985) also demonstrated the occurrence of cyclothymia among the adolescent offspring of bipolar adults. Furthermore, among monozygotic twins of patients with bipolar disorder who are not themselves affected, over half suffer from temperamental disturbances (Bertelsen et al., 1977). Thus, it could be assumed that temperamental attributes would represent predisposing familial-risk factors for affective illness. Further investigations regarding comorbidity, syndromal phenomenology, treatment response and temperament in relatives are now required to strengthen the hypothesis of a trait-state continuum between personality and affective episodes, which could both represent the expression of the same genetic or constitutional endowment.
Acknowledgements ˆ This work was supported by a grant from RhonePoulenc Rorer (JL) and Assistance Publique.
References Akiskal, H.S., 1983. The bipolar spectrum: New concepts in classification and diagnosis. In: L. Grinspoon (Ed.), Psychiatry Update: The American Psychiatric Association Annual Review. American Psychiatric Press, Washington, DC, pp. 271–292. Akiskal, H.S., 1984. Characterological manifestations of affective disorders. Toward a new conceptualization. Integr. Psychiatry 2, 83–88. Akiskal, H.S., 1992. The distinctive mixed states of bipolar I, II, and III. Clin. Neuropharmacol. 15, 632A. Akiskal, H.S., 1995. Developmental pathways to bipolarity: are juvenile-onset depressions pre-bipolar? J. Am. Acad. Child Adolesc. Psychiatry 34, 754–763.
108
C. Henry et al. / Journal of Affective Disorders 56 (1999) 103 – 108
Akiskal, H.S., Mallya, G., 1987. Criteria for the ‘soft’ bipolar spectrum: treatment implications. Psychopharmacol. Bull. 23, 68–73. Akiskal, H.S., Akiskal, K., 1992. Cyclothymic, hyperthymic and depressive temperaments as subaffective variants of mood disorders. In: Tasman, A. (Ed.), American Psychiatric Association Review. American Psychiatric Press, Washington, DC, pp. 43–62. Akiskal, H.S., Hirschfeld, R.M.A., Yerevanian, B.I., 1983. The relationship of personality to affective disorders: a critical review. Arch. Gen. Psychiatry 40, 801–810. Akiskal, H.S., Downs, J., Jordan, P., et al., 1985. Affective disorders in the referred children and younger sibling of manic depressives: mode of onset and prospective course. Arch. Gen. Psychiatry 42, 996–1007. Bech, P., Rafaelsen, O.J., Kramp, P., Bolwig, T.G., 1978. The Mania Rating Scale. Scale construction and inter-observer agreement. Neuropharmacology 17, 430–441. Bertelsen, A., Harvald, D., Hauge, M., 1977. A Danish twin study of manic depressive disorders. Br. J. Psychiatry 130, 330–351. Bourgeois, M.L., Hantouche, E.G., Akiskal, H.S., 1997. The EPIMAN and EPIDEP French studies of bipolarity. Preliminary results. J. Bipolar Disord. 1, 13–19. Cassano, G.B., Akiskal, H.S., Musetti, L., Perugi, G., Soriani, A., Mignani, V., 1989. Psychopathology, temperament, and past course in primary major depressions. Psychopathology 22, 278–288. Cassano, G.B., Akiskal, H.S., Perugi, G., Musetti, L., Savino, M., 1992. The importance of measures of affective temperaments in genetic studies of mood disorders. J. Psychiat. Res. 26, 257–268. Clayton, D., Hills, M., 1993. Choise and interpretation of models. In: Statistical models in epidemiology. Oxford University Press, Oxford, pp. 271–281. Dell’Osso, L., Placidi, G.F., Nassi, R., Freer, P., Cassano, G.B., Akiskal, H.S., 1991. The manic-depressive mixed state: familial, temperamental and psychopathologic characteristics in 108 female inpatients. Eur. Arch. Psychiatry Clin. Neurosci. 240, 234–239. Depue, R., Slater, J.F., Wolfstetter-Kausch, H., 1981. A behavioral paradigm for identifying persons at risk for bipolar depressive disorders: a conceptual framework and five validating studies. J. Abnorm. Psychol. 90, 381–437. Goodwin, F.K., Jamison, K.R., 1990. Manic-Depressive Illness. Oxford University Press, New York.
Hirschfeld, R.M.A., Klerman, G.L., 1979. Personality attributes and affective disorders. Am. J. Psychiatry 136, 67–70. Klein, D.N., Depue, R.A., Slater, J.F., 1985. Cyclothymia in the adolescent offspring of parents with bipolar affective disorder. J. Abnorm. Psychol. 94, 115–127. Klein, D.N., 1990. Depressive personality: reliability, validity, and relation to dysthymia. J. Abnorm. Psychol. 99, 412–421. Klein, D.N., Taylor, E.B., Harding, K., et al., 1988. Double depression and episodic major depression: demographic, clinical, familial, personality, and socioenvironmental characteristics and short-term outcome. Am. J. Psychiatry 145, 1226– 1231. Kraepelin, E., 1921. Manic-Depressive Insanity and Paranoia. Livingstone, Edinburgh. Kretschmer, E., 1936. Physique and Character (Miller, E., trans.). Kegan Paul (Eds.), London. Liebowitz, M.R., Stallone, F., Dunner, D.L., Fieve, R.R., 1979. Personality features of patients with primary affective disorder. Acta Psychiatr. Scand. 60, 214–224. Montgomery, S.A., Asberg, M., 1979. A new depressed scale designed to be sensitive to change. Br. J. Psychiatry 134, 382–389. Musetti, L., Perugi, G., Soriani, A., et al., 1989. Depression before and after age 65. A re-examination. Br. J. Psychiatry 155, 330–336. Norusis, M.J., 1992. SPPS Inc. SPSS for Windows 5.0.1. SPSS Inc., Chicago. Nurnberger, J.I., Blehar, M.C., Kaufman, C.A., York-Cooler, C., Simpson, S.G., Harkavy-Freedman, J., Severe, J.B., Malaspina, D., Reich, T., 1994. Diagnostic intreview for genetic studies. Arch. Gen. Psychiatry 51, 849–864. Perugi, G., Musetti, L., Simonini, E., Piagentini, F., Cassano, G.B., Akiskal, H.S., 1990. Gender-mediated clinical features of depressive illness: the importance of temperamental differences. Br. J. Psychiatry 157, 835–841. Schneider, K., 1958. Psychopathic Personalities. (Hamilton, M.W., trans.). Cassel, London. Strelau, J., 1983. Temperament, Personality and Arousal. Academic Press, London. Von Zerssen, D., Possl, J., 1990. The premorbid personality of patients with different subtypes of an affective illness. Statistical analysis of blind assignment of case history data to clinical diagnosis. J. Affect. Disord. 18, 39–50.