Journal of Psychosomatic Research, Vol. 46, No. 1, pp.

75–82, 1999
Copyright  1998 Elsevier Science Inc.
All rights reserved.
0022-3999/99 $–see front matter

S0022-3999(98)00053-1

PREVALENCE OF
ALEXITHYMIA AND ITS ASSOCIATION
WITH SOCIODEMOGRAPHIC VARIABLES
IN THE GENERAL POPULATION OF FINLAND

JOUKO K. SALMINEN,* SIMO SAARIJÄRVI,†

ERKKI ÄÄRELÄ,† TUULA TOIKKA* and JUSSI KAUHANEN‡

(Received 5 December 1997; accepted 8 April 1998)

Abstract—The prevalence of alexithymia and its association with sociodemographic variables were stud-
ied in a sample of 1285 subjects representing the general population of Finland. Alexithymia was mea-
sured with the 20-item Toronto Alexithymia Scale (TAS-20). Alexithymia was normally distributed in
the population in both genders, confirming that it is a personality dimension. The prevalence of alexi-
thymia was 13%. Men were alexithymic almost twice (17%) as often as women (10%). Multivariate anal-
ysis showed that alexithymia was associated with male gender, advanced age, low educational level, and
low socioeconomic status. As to the three factors of the TAS-20, men scored higher in factors 2 (difficulty
in describing feelings) and 3 (externally oriented thinking), but there was no gender difference in factor
1 (difficulty in identifying feelings). Comparative population studies in other countries are needed to find
out whether there are any differences in the prevalence of alexithymia between cultures.  1998 Else-
vier Science Inc.

Keywords: Alexithymia; Prevalence; Sociodemographic variables; Epidemiology; General population;

Personality.

INTRODUCTION

Alexithymia was originally defined by Sifneos in the 1970s as a constellation of per-

sonality features characterized by difficulties in the verbalization and expression of
affects and elaboration of fantasies, and use of action to express emotions [1]. These
features were first assumed to be typical of patients with “classical” psychosomatic
diseases. Later, Taylor and coworkers proposed that alexithymia reflects a deficit in
the cognitive processing of emotions or, more generally, a disturbance in the regula-
tion of emotions [2]. There is accumulating evidence that these features are preva-
lent both in medically and mentally ill persons [3, 4]. Alexithymia may also be pres-
ent in healthy populations [5–7]. It has been suggested that alexithymia is a
personality trait that is normally distributed in the general population [4]. Conse-
quently, alexithymia should be treated as a dimensional construct rather than a cat-

*Research and Development Centre of the Social Insurance Institution, Turku, Finland.
†Department of Psychiatry, Turku University Central Hospital, Turku, Finland.
‡Department of Public Health and General Practice, University of Kuopio, Kuopio, Finland.
Address correspondence to: Dr. Jouko K. Salminen, Research and Development Unit of the Social
Insurance Institution, Peltolantie 3, FIN-20720 Turku, Finland. Tel: 1358-2-2646411; Fax: 1358-2-
2646450; E-mail: [email protected]

75
76 J. K. SALMINEN et al.

egorical phenomenon. However, to facilitate comparisons between diverse popula-

tions, alexithymia is often treated as a categorical entity as well. The prevalence and
distribution of alexithymia have been widely studied in various clinical populations,
but few attempts have been made to study the association between alexithymia with
sociodemographic variables in the general population.
Epidemiological studies in nonclinical populations have mostly involved limited
and selected samples. A Finnish study with 266 subjects, drawn from the general
population and using the observer-rated Beth Israel Hospital Questionnaire (BIQ)
[1], showed that the prevalence of distinct alexithymia was rather low (4.1%),
whereas 21.8% of the subjects showed mild alexithymic features [8]. There was a
clear gender difference; men were more often alexithymic than women. Alexi-
thymia and alexithymic features were also associated with advanced age, lower so-
cial status, and singleness [8]. Being an observer-rated instrument, the BIQ is not
easily applicable in large population studies, especially in nonclinical circumstances.
The Toronto Alexithymia Scale (TAS-26), published in the mid-1980s was the
first psychometrically valid and reliable instrument for the assessment of alexi-
thymia [9]. This self-report questionnaire was developed further, resulting in the
20-item Toronto Alexithymia Scale (TAS-20) published in 1994 [10, 11]. This
is currently the best and most frequently used method in the measurement of
alexithymia. The TAS-20 contains three factors which reflect distinct facets
of alexithymia and that have been empirically validated [10–12]. The factors are: (1)
difficulty in identifying feelings; (2) difficulty in describing feelings; and (3) exter-
nally-oriented thinking.
Using the TAS-26, Pasini and coworkers found that in 417 normal subjects, aged
21–64 years, the total TAS-26 scores correlated positively with age and negatively
with education, but no gender difference was found [6]. No prevalence data were
presented. The largest epidemiological study so far has been the one carried out by
Kauhanen and coworkers [13]. They studied four cohorts of 2682 Finnish middle-
aged men drawn from the general population and found that income level and occu-
pational status were inversely related to the degree of alexithymia measured by the
TAS-26. Furthermore, men in the highest alexithymia quartile were almost four
times as likely to be unmarried as men in the lowest quartile. No prevalence data
could be presented due to a minor modification of the TAS-26 used in this study.
In a nonclinical elderly population (mean age 72 years), Joukamaa et al. found the
prevalence of alexithymia, measured by the TAS-26, to be 34%. Alexithymia was as-
sociated with poor perceived somatic health, but not with gender, marital status, social
status, or residential area [7]. In sum, the few studies on alexithymia in normal pop-
ulations and on its association with sociodemographic variables are controversial.
In this study we investigate the prevalence of alexithymia in the general popula-
tion of Finland and its association with sociodemographic variables.

METHOD
Subjects
The population comprised 2000 persons of working age (18–64 years), randomly sampled from the
population register of the Social Insurance Institution in Finland. The sample was stratified according
to the five social insurance districts over all of Finland, and the sample sizes were proportional to the
size of the population in each district. One subject died before the data collection.
 Alexithymia in the Finnish population 77
The TAS-20 used in this study had been translated into Finnish and retranslated into English by J.
Julkunen and J. Kauhanen in 1994 with permission from G. Taylor. The accuracy of the retranslation
was checked by a native speaker of English with academic education. This translation is used by the re-
searchers in Finland.
A self-administered questionnaire including sociodemographic variables and the TAS-20 was mailed
in August 1996 to 978 men and 1021 women. Another questionnaire was sent to nonresponders 4 weeks
later. A total of 1336 persons (66.8%) returned the questionnaire; 597 (44.7%) of them were men with
mean age of 39.8 (sd 13.1) years, and 739 (55.3%) were women with mean age of 39.3 (sd 13.1) years.
No differences in gender (p50.52) or age (p50.40) were found between the social insurance districts.
The remaining 663 persons, 381 (57.5%) males and 282 (42.5%) females, did not return the question-
naire. The proportion of men was higher in the nonresponding group than in the study group (p,0.001).
The nonresponding men were slightly younger, mean age 37.1 (sd 12.0) years, than those responding
(p,0.001). In women, there was no difference in mean age between the nonresponders, 38.9 (sd 11.9)
years, and the study group, 39.3 (sd 13.1) (p50.63).
The analyses of alexithymia are based on 1285 persons, 577 men (44.9%) and 708 women (55.1%),
who had complete data in the TAS-20 questionnaire.

Procedure
To assess the prevalence of alexithymia, the TAS-20 total scores were categorized according to the
cutoff points suggested by the Toronto group: a total score of >61 points indicates alexithymia, and <51
points indicates no alexithymia [4]. The total score of the TAS-20 was compared separately with each
of the following sociodemographic variables: age, gender, educational level, socioeconomic status, and
marital status. The educational level was dichotomized using the matriculation examination as criterion:
secondary school graduates vs. nongraduates. Socioeconomic status was determined according to a Finn-
ish statistical handbook [14], which takes into account the effect of education. For marital status, married
or cohabiting persons formed one group and single persons (i.e., widows, divorced, and unmarried) the
other group. The gender difference in the three factors of the TAS-20 was also tested.

Data analysis
The SAS program package, version 6.09 [15], was used for all statistical analyses. Pearson’s chi-
square test was used for between-group comparisons of categorical variables. For numerical variables,
Student’s t-test was used in pairwise comparisons between groups and analysis of variance between three
or more groups (Bonferroni probabilities). Alexithymia (with TAS-20 score as a numerical variable) was
analyzed with linear models using the GLM (general linear models) procedure in which the independent
variables were both numerical and categorical. All tests were two-sided.

RESULTS

The sociodemographic characteristics of the study group are presented in Table

I. There was no difference between men and women in age (p50.32) or in marital
status (p50.31). There were more secondary school graduates among women than
among men (p,0.001). In regard to socioeconomic status, there was a difference be-
tween genders (p,0.001), with men more often being private enterpreneurs or
blue-collar workers than women, and women being more often white-collar work-
ers than men.

Prevalence of alexithymia
The overall prevalence of alexithymia (TAS-20 score >61) was 12.8%. The prev-
alence was 16.6% among men and 9.6% among women (p,0.001). No alexithymia
(TAS-20 score <51) was found in 68.0% of the study group; 59.8% of men and
74.7% of women were not alexithymic.
The overall mean score of the TAS-20 for the whole population was 46.0 (sd
11.6). The scores were normally distributed in both genders, but there was a signifi-
cant gender difference in the mean scores, men being more alexithymic than women
78 J. K. SALMINEN et al.

Table I.—Sociodemographic variables of the study groupa

Men Women
(n 5 577, 45% (n 5 708, 55% Total
of patients) of patients) (n 5 1285)

Mean (sd) age 39.6 (13.0) 38.9 (13.0) 39.2 (13.0)

Educational level
Secondary school
graduate 163 (28.6) 266 (37.9) 429 (33.8)
Nongraduate 407 (71.4) 435 (62.1) 842 (66.2)
Marital status
Married 375 (65.7) 479 (68.1) 854 (67.0)
Single 202 (35.0) 229 (32.3) 431 (33.5)
Socioeconomic status
Private entrepreneur 74 (12.8) 48 (6.8) 122 (9.5)
White-collar worker 190 (32.9) 363 (51.3) 553 (43.0)
Blue-collar worker 236 (40.9) 188 (26.5) 424 (33.0)
Others 77 (13.4) 109 (15.4) 186 (14.5)
a
Data expressed as mean, with percentage in parentheses.

(Fig. 1). The mean of the total TAS-20 score was 48.8 (sd 11.4) for men and 43.8
(sd 11.3) for women (p,0.001).
As to the three factors of the TAS-20 the gender difference was observed in fac-
tor 2 with a mean of 12.7 (sd 4.3) for men and 10.7 (sd 4.0) for women (p,0.001),

Fig. 1. Distribution of the TAS-20 scores shown by box-and-whisker plots.

 Alexithymia in the Finnish population 79

and in factor 3 with a mean of 21.8 (sd 4.6) for men, and 18.7 (sd 4.8) for women
(p,0.001). There was no difference in factor 1 between men, mean 14.3 (sd 5.6),
and women, mean 14.4 (sd 5.3), (p50.68).

Sociodemographic variables associated with alexithymia

Different models fitted into the data showed that there were no interactions be-
tween gender and other sociodemographic variables. When women and men were
tested separately, it was found that, in both genders, secondary school graduates
were less alexithymic than those without matriculation examination (p,0.001).
Married women were less alexithymic than single women (p,0.01), but in men
there was no association between marital status and alexithymia. Blue-collar work-
ers were more alexithymic than white-collar workers in both genders (p,0.001 in
men and p,0.01 in women). There was a weak positive correlation between ad-
vanced age and TAS-20 score (r50.15 [p,0.001] in men, r50.10 [p,0.001] in
women).
The TAS-20 scores by educational level, marital status, and socioeconomic status
are presented in Table II.
Multivariate GLM model analysis (n51271) showed significant independent as-
sociations between alexithymia (total score of the TAS-20) and the following socio-
demographic variables: male gender (p,0.001), low educational level (p,0.001),
low socioeconomic status (p,0.001), and advanced age (p50.004). Marital status
was not significantly associated with alexithymia.

DISCUSSION

This is the first study on the prevalence of alexithymia in the general population
using the TAS-20 in a representative sample. The response rate of 67% is accept-
able. There were proportionally more women and less young men among the re-
sponders than in the nonresponders. In spite of these limitations the sample repre-
sents fairly well the Finnish general population. Our study confirmed that
alexithymia, measured by the TAS-20, is normally distributed in the population in
both genders. This supports the idea that alexithymia is a dimensional personality
trait [4, 16].
The total prevalence of alexithymia, as measured by the TAS-20, was 13% (17%
in men and 10% in women). These prevalence figures are about three times as high
as those of an earlier Finnish study using the BIQ [8]. The cutoff points for the BIQ
are not empirically established, which complicates the comparison of the prevalence
rates of the two studies.
Our study showed a distinct gender difference in the prevalence of alexithymia:
men were alexithymic almost twice as often as women. A similar gender difference
was found in the earlier Finnish study using the BIQ [8], which, to our knowledge,
is the only population-based study on alexithymia in both genders preceding our
study. The study by Parker et al. [12] using the TAS-20, and covering three samples
of young adults in Canada, the United States, and Germany, showed similar corre-
lations between alexithymia and gender as those found in our study. On the other
hand, the Italian study by Pasini and coworkers [6] using the TAS-26 showed no
gender difference in alexithymia, but it must be pointed out that their study group
 80

Table II.—The TAS-20 scores by educational level, marital status and socioeconomic status

Men (n 5 577) Women (n 5 708) Total (n 5 1285)

n Mean (sd) p2b n Mean (sd) n Mean (sd)

Educational level
Secondary school graduate 163 43.5 (9.3) p , 0.001 266 39.6 (10.0) 429 41.1 (9.9)
Nongraduate 407 50.9 (11.5) p , 0.001 435 46.2 (11.3) 842 48.5 (11.6)
p1a p , 0.001 p , 0.001 p , 0.001
Marital status
Married 375 48.9 (11.4) p , 0.001 479 43.0 (11.0) 854 45.6 (11.5)
Single 202 48.6 (11.4) 0.004 229 45.3 (11.9) 431 46.9 (11.8)
p1a 0.78 0.01 0.06
Socioeconomic status
Private entrepreneur 74 48.6 (11.0) 0.03 48 43.9 (12.4) 122 46.7 (11.8)
White-collar worker 190 45.5 (8.8) p , 0.001 363 42.2 (10.6) 553 43.3 (10.4)
J. K. SALMINEN et al.

Blue-collar worker 236 51.9 (12.1) p , 0.001 188 45.8 (12.1) 424 49.2 (12.4)
Others 77 47.6 (11.2) 0.24 109 45.6 (11.4) 186 46.4 (11.3)
p , 0.001c p , 0.001d p , 0.001e
a
Differences within group.
b
Differences between genders.
c
Differences between blue-collar and white-collar workers p , 0.001; differences between blue-collar workers and others p , 0.05.
d
Differences between white-collar and blue-collar workers p , 0.01; differences between white-collar workers and others p , 0.05.
e
Differences between white-collar workers and private entrepreneurs p , 0.01; differences between white-collar and blue-collar workers p , 0.001;
differences between white-collar workers and others p , 0.01.
 Alexithymia in the Finnish population 81

consisted of healthy volunteers. It seems likely that alexithymia is a personality trait

that is more typical of men than women.
There were differences between men and women in regard to the different facets
of the alexithymia construct, represented by the three factors of the TAS-20. Men
and women were equally capable of identifying feelings (factor 1), but women were
more able to describe their feelings (factor 2) and less prone to externally oriented
thinking (factor 3) than men. The gender differences in the two latter factors of the
TAS-20 (i.e., that women are more able to express and describe their feelings and
that their thinking is more internally oriented as compared with men) could be ex-
plained by gender-related differences in early learning experiences that reflect cul-
turally held ideals of coping with emotions. That no gender difference was observed
in the first factor (i.e., in identifying feelings) is a new and interesting finding that
needs further investigation.
Both our study and the study by Lindholm et al. [8] showed that alexithymia is
associated with advanced age. This association has often been interpreted as a sec-
ondary phenomenon due to poor physical health of elderly people [7, 17, 18]. When
falling ill, people may become emotionally more constricted [18–20]. Another ex-
planation is that elderly people have grown up in a different cultural environment
and may have learned different ways to deal with their affects as compared with
younger generations.
Previous Finnish population-based studies [8, 13] have shown an association be-
tween alexithymia and singleness. In our study, married women were less alexi-
thymic than single women. However, in the multivariate analysis this association
disappeared, showing that it is explained by other factors.
The association between alexithymia and low educational level was quite distinct
in our study. This finding contrasts with the results of Parker and coworkers in Can-
ada, who found no association with educational level and alexithymia [5]. The most
probable explanation for this discrepancy is that the subjects in the Canadian study
did not represent the general population, because they consisted of a small group
of volunteers recruited from passenger lounges at a railway station and an airport.
The subjects in earlier studies showing no association between educational level and
alexithymia have often been college or university students, and therefore conclu-
sions drawn in these studies cannot be generalized to the whole population. It seems
plausible that alexithymic persons are less likely to seek higher education. The edu-
cational level reflects the social status, values, and emotional atmosphere in the
family of origin, thus yielding information about the individual’s developmental
background. Further developmental and longitudinal studies of alexithymic persons
are needed to address these issues.
In the present study, we found that alexithymia is associated with low socioeco-
nomic status. This is in accordance with the earlier Finnish population studies [8,
13]. Development of affects and of imaginative capacity in childhood is influenced
by the child’s familial environment which, in turn reflects the parents’ educational
level and socioeconomic status [13]. As a personality trait, alexithymia may have a
negative influence on an individual’s social success, resulting in lower socioeco-
nomic status later in life. An alternative explanation is that actual socioeconomic
difficulties influence an individual’s ability to deal with emotions.
In conclusion, alexithymia is a personality dimension that is normally distributed
in the general population. Alexithymic personality traits were associated with male
82 J. K. SALMINEN et al.

gender, advanced age, low educational level, and low socioeconomic status, but not
with marital status. In addition to its social associations, alexithymia may have im-
portant connections with individual health status. It may be one link in the chain
leading to the accumulation of socioeconomic deprivation and poor health in cer-
tain segments of the population. In fact, association of alexithymia with increased
mortality in middle aged men has been reported by Kauhanen et al. [21]. Future
studies will show to what extent alexithymia increases susceptibility to diseases in
general. Comparative population studies in other countries are needed to deter-
mine whether there are any differences in the prevalence of alexithymia between
various cultures.

REFERENCES

1. Sifneos PE. The prevalence of ‘alexithymic’ characteristics in psychosomatic patients. Psychother

Psychosom 1973;22:255–262.
2. Taylor GJ, Bagby RM, Parker JDA. The alexithymia construct. A potential paradigm for psychoso-
matic medicine. Psychosomatics 1991; 32;153–164.
3. Salminen JK, Saarijärvi S, Äärelä E. Two decades of alexithymia. J Psychosom Res 1995;39:803–807.
4. Taylor GJ, Bagby RM, Parker JDA. Disorders of affect regulation. Cambridge: Cambridge Univer-
sity Press 1997.
5. Parker JDA, Taylor GJ, Bagby RM. The alexithymia construct: relationship with sociodemographic
variables and intelligence. Compr Psychiatry 1989;30:434–441.
6. Pasini A, Chiaie RD, Seripa S, Ciani N. Alexithymia as related to sex, age, and educational level:
results of the Toronto Alexithymia Scale in 417 normal subjects. Compr Psychiatry 1992; 33:42–46.
7. Joukamaa M, Saarijärvi S, Muuriaisniemi ML, et al. Alexithymia in normal elderly population. Com-
preh Psychiatry 1996;37:144–147.
8. Lindholm T. Lehtinen V, Hyyppä MT, et al. Alexithymic features in relation to the dexamethasone
suppression test in a Finnish population sample. Am J Psychiatry 1990;147:1216–1219.
9. Taylor GJ, Ryan DP, Bagby RM. Toward the development of a new self-report alexithymia scale.
Psychother Psychosom 1985;44:191–199.
10. Bagby RM, Parker JDA, Taylor GJ. The twenty-item Toronto Alexithymia Scale—I. Item selection
and cross validation of the factor structure. J Psychosom Res 1994;38:23–32.
11. Bagby RM, Taylor GJ, Parker JDA. The twenty-item Toronto Alexithymia Scale—II. Convergent,
discriminant, and concurrent validity. J Psychosom Res 1994;38:33–40.
12. Parker JDA, Bagby RM, Taylor GK, et al. Factorial validity of the 20-item Toronto Alexithymia
Scale. Eur J Pers 1993;7:221–232.
13. Kauhanen J, Kaplan GA, Julkunen J, et al. Social factors in alexithymia. Compr Psychiatry 1993;
34:330–335.
14. The classification of socio-economic groups. Handbook of the Central Statistical Office of Finland
[in Finnish with an English summary]. Helsinki; CSO 1989.
15. SAS Institute. SAS/STAT user’s guide, version 6, 4th ed, vols 1 and 2. Cary, North Carolina: SAS
Institute 1990.
16. Salminen JK, Saarijärvi S, Äärelä E, et al. Alexithymia—state or trait? One-year follow-up study of
general hospital psychiatric consultation out-patients. J Psychosom Res 1994;38:681–685.
17. Smith GR. Alexithymia in medical patients referred to a consultation-liaison service. Am J Psychia-
try 1983;140:99–101.
18. Wise TN, Mann LS, Mitchell JD, et al. Secondary alexithymia: an empirical validation. Compr Psy-
chiatry 1990;31:284–288.
19. Freyberger H. Supportive psychotherapeutic techniques in primary and secondary alexithymia. Psy-
chother Psychosom 1987;28:337–342.
20. Keltikangas-Järvinen L. Concept of alexithymia, II: The consistency of alexithymia. Psychother Psy-
chosom 1987;47:113–120.
21. Kauhanen J, Kaplan GA, Cohen RD, et al. Alexithymia and risk of death in middle aged men. J
Psychosom Res 1996;41:541–549.