Am J Psychiatry 161:7, July 2004 1299 http://ajp.psychiatryonline.org Brief Report Alexithymia and Personality in Relation to Dimensions of Psychopathology Hans Joergen Grabe, M.D. Carsten Spitzer, M.D. Harald Juergen Freyberger, M.D. Objective: The authors examined the capacity of alexithymia to predict a broad range of psychiatric symptoms relative to that of other personality dimensions, age, and gender. Method: The Toronto Alexithymia Scale, the Temperament and Character Inventory, and the SCL-90-R were administered to 254 psychiatric patients. Multivariate linear regression analy- ses were performed. Results: The difficulties identifying feelings factor of the Toronto Alexithymia Scale significantly predicted all SCL-90-R subscale scores and was particularly effective, relative to the personality dimensions of the Temperament and Character Inventory, in pre- dicting somatization. The Temperament and Character Inventory dimensions emerged as distinct and conceptually meaningful predictors for the different SCL-90-R subscales. Conclusions: A broad range of current psychopathology is asso- ciated with difficulties in cognitively processing emotional per- ceptions. Further research needs to clarify whether alexithymia represents a risk factor for mental illness and poorer outcome. (Am J Psychiatry 2004; 161:1299–1301) Most psychotherapeutic approaches aim to in- crease the perceptual awareness of emotional processes, to enhance emotional differentiation, and to enable— through insight or cognitive behavioral strategies—new emotional experiences to help overcome dysfunctional mental states. These emotional deficits may be in fact de- scribed as alexithymia (1), focusing primarily on distur- bances of affect recognition, affect differentiation, and on the expression of feelings. Although alexithymia was first studied in a variety of medical and psychosomatic dis- orders (1, 2), more recent studies have associated alexi- thymia with dissociation (3), depression (1, 4, 5), anxiety disorders, hypochondria, and eating disorders (1). How- ever, the contribution of alexithymia in interaction with common personality traits to psychopathology has not been investigated systematically. This is important because alexithymia is captured up to 50% by personality models like the five-factor model of personality or Cloninger’s psychobiological model of personality (6, 7). We hypothesized that disturbances in identifying and differentiating feelings would predict a broad range of psychopathology. Method The study group consisted of 254 randomly selected, postacute inpatients and outpatients (156 women and 98 men). All patients were Caucasian and gave written informed consent after receiv- ing full explanation of the study. Patient diagnoses were made ac- cording to DSM-IV. The study group comprised subjects with de- pressive disorders (N=78), anxiety disorders (N=34), obsessive- compulsive disorder (N=28), schizophrenia and schizoaffective disorder (N=32), borderline personality disorder (N=36), dissocia- tive disorder (N=10), mixed personality disorder (N=18), alcohol dependency (N=9), and somatization disorder (N=9). Alexithymia was assessed with the German-language transla- tion of the 20-item Toronto Alexithymia Scale (8–10), which com- prised three factors: 1) difficulties identifying feelings, 2) difficul- ties expressing feelings, and 3) externally orientated thinking. Personality was assessed with the Temperament and Character Inventory (11, 12), a 240-item, forced-choice, self-report scale. The SCL-90-R (13), a 90-item self-report scale, was used to assess current psychopathology on nine subscales. All reported scores provided dimensional information, with higher scores indicating a higher degree of alexithymia, a more extreme expression of per- sonality traits, and poorer psychopathology, respectively. Nine linear regression analyses were calculated using the sub- scores of the SCL-90-R as dependent variables. The Temperament and Character Inventory dimensions, the Toronto Alexithymia Scale factors, age, and gender were entered blockwise (multivari- ate) into each of the nine equations as independent variables. The significance of the explained amount of variance (R 2 ) by the re- gression equation compared with the total variance of the depen- dent variable was assessed by using analysis of variance. For each independent variable, a standardized regression coefficient (β) is reported, which is based on a prior z transformation of the inde- pendent variables. Therefore, β indicates the relative magnitude of prediction of each independent variable. The significance of β was evaluated by a t statistic. All predictor variables had tolerance values >0.4 excluding collinearity. Results High scores on difficulties identifying feelings emerged as a major predictor for current psychopathology on all SCL-90-R subscales and were particularly effective in predicting “somatization” (Table 1). Except for a weak pre- diction of “phobia” by low externally orientated thinking, none of the SCL-90-R subscales were predicted by difficul- ties expressing feelings or externally orientated thinking. Gender (male subjects) only emerged as a weak predictor for “aggression.”