TEMPS-I: delineating the most discriminant traits of the cyclothymic, depressive, hyperthymic and irritable temperaments in a nonpatient population

https://doi.org/10.1016/S0165-0327(98)00152-9Get rights and content

Abstract

Background: Although most personality constructs have been standardized in population studies, cyclothymic, depressive, irritable and hyperthymic temperaments putatively linked to mood disorders have been classically derived from clinical observations. Methods: We therefore administered the semi-structured affective temperament schedule of Memphis, Pisa, Paris and San Diego, Interview version (TEMPS-I) — in its original University of Tennessee operationalization — to 1010 Italian students aged between 14 and 26. The interview, administered in a randomized format, took 20 min per subject. Results: The semi-structured interview was easy to administer and well accepted by subjects, with no refusals. Principal component analysis with varimax rotation confirmed the hypothesized four-dimensional factor structure of the interview, with good to excellent internal consistency. Furthermore, discriminant analysis and multiple regression provided suggestions for identifying the traits that are most useful in defining a weighted cut-off for each of the temperaments (and which, with minor exceptions, are in agreement with those previously proposed on clinical grounds). In an additional exploratory factorial analysis, a depressive type which loads negatively on hyperthymia was distinguished from cyclothymia; the irritable temperament did not appear to have significant loading on either factor. Limitation: All the present analyses were internal to the scale itself, but ongoing studies are comparing them with other systems of temperament as well as testing their clinical cogency for affectively ill populations. Conclusion: While more work needs to be done on better operationalization of the irritable temperament, our findings overall support the existence — in a relatively young nonpatient population — of cyclothymic, depressive and hyperthymic types according to the classic descriptions of Kraepelin, Kretschmer and Schneider, in their TEMPS-I operationalization. Clinical implications: Coupled with a previous report identifying 10% of the same 14–26-year-old nonpatient population meeting an empirically defined statistical cut-off for these temperaments, the present data define the putative `fundamental states' that Kraepelin considered to be the personal predisposing anlagé of major affective disorders.

Introduction

The criteria for affective temperaments developed at the University of Tennessee (Akiskal et al., 1977, Akiskal et al., 1979) both in their original form or modified versions, have been widely used to detect the temperamental attributes of clinical (Perugi et al., 1990, Klein, 1990, Koukopulos et al., 1990) and nonclinical populations (Depue et al., 1981, Klein and Miller, 1993). However, no studies have specifically tested the validity of the entire set of the cyclothymic, irritable, depressive and hyperthymic temperaments administered in semi-structured format (Akiskal and Mallya, 1987). The operationalization of these constructs began with our study on cyclothymia, conducted at the University of Tennessee mood clinic (Akiskal et al., 1977). The definitions of the hyperthymic and depressive types (Akiskal et al., 1979) benefited greatly from the perspicacity and astute clinical observations of our late teacher and Memphis colleague Professor Vahé Puzantian, who was well-grounded in the classical German descriptions (Kraepelin, 1921, Schneider, 1958). Thus, the cyclothymic, hyperthymic and depressive types had always been part of the Mood Clinic Data Questionnaire (Akiskal et al., 1978), used in most of the research emanating from that Memphis-based clinic. An irritable type was formally added a decade later (Akiskal and Mallya, 1987).

A brief review of research conducted prior to our Italian collaboration will better set the stage for the analyses to be reported in this paper. In a study on a large sample of college students, Depue et al. (1981)indicated that the most frequent items for cyclothymia —in large part derived from Akiskal et al., 1977, Akiskal et al., 1979— were `lethargy alternating with eutonia', `marked unevenness in quantity and quality of productivity' and `mental confusion alternating with sharpened and creative thinking,' while the least frequent were `alternation between low self-confidence and overconfidence' and `decreased verbal output alternating with talkativeness.' In an affectively ill population, Cassano et al. (1989), using explicitly two of the temperaments operationalized by Akiskal and Mallya (1987), reported that the most frequent traits observed in clinically depressed patients with hyperthymic temperament were `irritable, cheerful, overoptimistic or exuberant, `warm, people-seeking or extroverted' and `overtalkative,' while the least frequent were `overinvolved and meddlesome,' `uninhibited, stimulus-seeking or promiscuous'; the depressive temperament showed a more homogeneous profile and the most frequent traits, again in a clinically depressed population, were `brooding and given to worry,' `conscientious or self-disciplining,' and `quiet, passive or indecisive,' while the least frequent were `preoccupied with inadequacy, failure and negative events to the point of morbid enjoyment of one's failures' and `sceptical, hypercritical or complaining.'

Both of the foregoing studies reported the frequency of single items without permitting an assessment of the weight and predictive value of each item or identification of the most discriminant traits. In fact, the most frequent items analyzed by means of the univariate method do not always prove to be the most discriminant. To our knowledge, only two studies (Klein, 1990, Klein and Miller, 1993) have evaluated the reliability and validity of the depressive temperament construct of Akiskal and Mallya. By using the Akiskal–Mallya depressive personality item set in college student and outpatient groups, these authors demonstrated that the four items most closely related to the overall construct were `gloomy, pessimistic, serious, incapable of fun,' `self-critical, self-reproaching and self-derogatory, `brooding and given to worry' and `preoccupied with negative events, feelings of inadequacy.' On the other hand, the items `quiet, introverted, passive, and nonassertive,' `sceptical, critical of others,' `conscientious, responsible and self-disciplined' had the lowest item-total correlations.

In a previous paper (Placidi et al., 1998), the standardization of the Semi-Structured Affective Temperament Interview — originally developed in the University of Tennessee at Memphis and under current evaluation in Pisa, Paris and San Diego (TEMPS-I) — made it possible to compare the three scales of depressive, hyperthymic and cyclothymic temperaments by showing the dominant temperamental characteristics for each subject. In the present report we carried out a different set of statistical analyses in our sample in order to identify the most discriminant traits of the four affective temperament scales in their University of Tennessee operationalization and, in a more general way, to confirm the validity of these constructs deriving from classic German psychiatry.

Section snippets

Subjects

The characteristics of the subjects and the instruments used in this study have been reported in detail in a previous paper (Placidi et al., 1998). The randomized student sample consisted of 1010 subjects (n=518 males, n=492 females; mean age of 18±3 years, and range of 14 to 26). The randomized sample was drawn from a student population attending five schools (n=704) providing secondary education with three different orientations (technical, professional and humanities), and taking various

Results

Table 2 shows the data resulting from the confirmatory factor analysis of the item scores of TEMPS-I. The hypothesized factorial structure in four dimensions has been confirmed; the cyclothymic factor appears to explain the highest proportion of variance, followed in decreasing order by the irritable, depressive and hyperthymic factors. It is particularly noteworthy that depressive factor correlates negatively with items belonging to the hyperthymic construct (HYP-1 and HYP-2): in other words,

Discussion

The data emerging from the multivariate analysis of TEMPS-I administered to 1010 young subjects further confirm the internal validity of this construct. In fact, confirmatory factor analysis with varimax rotation provides four factors, thereby ruling out other dimensions in the temperament affective scales. Moreover, all but two traits (DEP-3 `sceptical, hypercritical or complaining' and IRR-3 `tendency to brood' included, respectively, in the depressive and irritable scales) show a good

Conclusions

Multivariate analysis makes it possible to confirm the construct validity of the Akiskal and Mallya Semi-Structured Interview for the Affective Temperaments. Four temperaments emerge, of which the irritable is the least well defined. Of the irritable temperament, we recommend retaining only items IRR-2, IRR-3, IRR-5 and IRR-8 for further systematic study. Of the depressive, DEP-3 could be provisionally deleted. Of the cyclothymic, we recommend dropping CYC-6 and CYC-10. The hyperthymic scale

References (33)

  • H.S. Akiskal et al.

    Criteria for the `soft' bipolar spectrum: Treatment implications

    Psychopharmacol. Bull.

    (1987)
  • H.S. Akiskal

    Delineating irritable-choleric and hyperthymic temperaments as variants of cyclothymia

    J. Person. Disord.

    (1992)
  • H.S. Akiskal

    Toward a temperament-based approach to depression: Implications for neurobiologic research

    Adv. Biochem. Psychopharmacol.

    (1995)
  • G.B. Cassano et al.

    Psychopathology, temperament, and past course in primary major depression, 2: toward a redefinition of bipolarity with a new semi-structured Interview for depression

    Psychopathology

    (1989)
  • R.A. Depue et al.

    A behavioral paradigm for identifying persons at risk for bipolar depressive disorders: A conceptual framework and five validation studies

    J. Abnorm. Psychol.

    (1981)
  • M. Eckbald et al.

    Development and validation of a scale for hypomanic persona

    J. Abnorm. Psychol.

    (1986)
  • Cited by (0)

    1

    University of Tennessee, Memphis (deceased).

    View full text