Alexithymia as Predictor of Treatment Outcome in Patients with Functional

Gastrointestinal Disorders
PIERO PORCELLI, PHD, R. MICHAEL BAGBY, PHD, GRAEME J. TAYLOR, MD, FRCPC, MASSIMO DE CARNE, MD,
GIOACCHINO LEANDRO, MD, AND ORLANDO TODARELLO, MD
Objective: A previous study found a strong association between alexithymia and functional gastrointestinal disorders (FGID). The
objective of this study was to investigate whether alexithymia might be a predictor of treatment outcome in patients with FGID.
Methods: A group of FGID outpatients classified by the Rome I criteria was divided into improved (N 68) and unimproved
(N 44) groups on the basis of pre-established criteria after 6 months of treatment. Patients were administered the 20-item Toronto
Alexithymia Scale, the Hospital Anxiety and Depression Scale, and the Gastrointestinal Symptom Rating Scale both before and
after 6 months of treatment. Results: At the base-line assessment, compared with the improved patients, the unimproved patients
had significantly higher levels of anxiety, depression, alexithymia, and gastrointestinal symptoms. Stability of alexithymia was
demonstrated by significant correlations between base-line and follow-up TAS-20 scores in the entire sample. Moreover,
hierarchical regression analyses showed that the stability of TAS-20 scores over the 6-month treatment period could not be
accounted for by their associations with anxiety and depression scores. In logistic regression analyses, base-line alexithymia and
depression emerged as significant predictors of treatment outcome. Relative to depression, however, alexithymia was the stronger
predictor. Conclusions: Alexithymia is a reliable and stable predictor of treatment outcome in FGID patients. Although further
studies are needed, clinicians might improve treatment outcome by identifying patients with high alexithymia, and attempting to
improve these patients skills for coping with emotionally stressful situations. Key words: alexithymia, functional gastrointestinal
disorders, anxiety, depression, prediction, treatment outcome.
FGID functional gastrointestinal disorders; GSRS Gastrointestinal Symptom Rating Scale; HADS Hospital Anxiety and Depression Scale; HADS-A Hospital Anxiety and Depression Scale,
Anxiety subscale; HADS-D Hospital Anxiety and Depression
Scale, Depression subscale; TAS-20 Twenty-item Toronto Alexithymia Scale.

INTRODUCTION
unctional gastrointestinal disorders (FGID) occur commonly in the general population and comprise a high percentage of patients who present to primary care physicians or
gastroenterologists (1 4). Although the pathophysiology of
FGID is not fully understood, there is accumulating evidence
that symptom formation involves an interaction among multiple factors that vary in importance from one patient to
another. These factors include motility disturbances, altered
thresholds of pain and other sensory input from the gut,
gastrointestinal inflammation and infection, psychological distress, and personality disturbances (510). Personality traits
and emotional states may not only induce effects on the
physiology of the gut (6, 11), but also influence how the
symptoms of FGID are experienced and acted on, and the
outcome of treatment (9, 1213).
One personality trait that has been associated with both
illness behavior and emotion dysregulation is alexithymia (14,
15). As defined by Nemiah et al. (16), the salient features of
the alexithymia construct are: (a) difficulty identifying feel-

From the Psychosomatic Unit (P.P.) and Department of Gastroenterology,

IRCCS De Bellis Hospital (M.D.C., G.L.), Castellana Grotte, Italy; Department of Psychiatry, University of Toronto and Centre for Addiction and
Mental Health-Clarke Site (R.M.B.), Toronto, Ontario, Canada; Department
of Psychiatry, University of Toronto and Mount Sinai Hospital (G.J.T.),
Toronto, Ontario, Canada; and Department of Psychiatry, University of Bari
(O.T.), Italy
Address reprint requests to: Dr. Piero Porcelli, Unit di Psicosomatica,
IRCCS Ospedale De Bellis, Via Valente, 4, I-70013, Castellana Grotte
(Bari), Italy. E-mail: porcellip@mail.media.it
Received for publication August 15, 2002; revision received November 25,
2002.
DOI: 10.1097/01.PSY.0000089064.13681.3B
Psychosomatic Medicine 65:911918 (2003)
0033-3174/03/6505-0911
Copyright 2003 by the American Psychosomatic Society

ings and distinguishing between feelings and the bodily sensations of emotional arousal; (b) difficulty describing feelings
to others; (c) constricted imaginative processes; and (d) a
stimulus-bound, externally orientated cognitive style. These
characteristics are thought to reflect deficits in the cognitive
processing and regulation of emotions (14, 15). In a preliminary study, a group of patients with FGID, classified by the
Rome I criteria (5), was found to be significantly more
alexithymic than a group of patients with inflammatory bowel
disease, and the two groups were significantly more alexithymic than a group of healthy adults (17). Moreover, 66% of the
FGID patients scored in the high range (61) of the 20-item
Toronto Alexithymia Scale (TAS-20). This study used a crosssectional design and therefore provided no information about
the stability of alexithymia or its impact on the course of
FGID and patients responses to treatment.
In assessing the stability of a personality trait, such as
alexithymia, it is important to distinguish between absolute
stability and relative stability. Absolute stability refers to the
extent to which personality test scores in an identified group
of individuals, as a whole, change over time, including during
the course of treatment. Relative stability refers to the degree
to which the relative differences on personality test scores
among individuals in a group remain the same over time, even
in the context of acute symptom change (18). Although several investigators have concluded that alexithymia is unstable
on the basis of a failure to demonstrate absolute stability,
Luminet et al. (19) have argued that relative stability, even in
the absence of absolute stability, is sufficient to establish
alexithymia as a vulnerability factor for certain medical and
psychiatric disorders, and as a moderator of their response to
treatment. The aim of the present study was to assess the
stability of alexithymia in patients with FGID and to investigate the extent to which alexithymia can predict the response
to the treatment strategies that are usually administered to
FGID patients.
911

P. PORCELLI et al.
METHODS
Participants and Procedures
The participants were 130 consecutive patients (44 men and 86 women)
referred for their first consultation at the FGID outpatient facility of a
gastrointestinal tertiary care hospital in southern Italy, between January 1997
and December 1998. All patients satisfied the Rome I symptom-based
diagnostic criteria for one or more of the FGID subgroups (functional dyspepsia, irritable bowel syndrome, or functional abdominal pain) (5), and for
all patients, biochemical, ultrasonographic, and endoscopic examinations
revealed no evidence of any structural organic gastrointestinal disease. The
mean age of the group was 37.9 years (SD 14.4) and the mean education
was 9.5 years (SD 3.8). All of the patients were willing to participate in the
study and gave written informed consent.
At the time of their initial evaluation (base-line) all patients completed
self-report psychological scales and were rated by their gastrointestinal physician on a gastrointestinal symptom questionnaire (see Measures section).
Patients were treated according to their symptoms case by case with combination forms of gastrointestinal medications (usually antisecretory, prokinetic,
and antispasmodic drugs), diet modifications (usually including high-fiber
intake), psychotropic medications (usually anxiolytic and antidepressant
drugs, in low doses), and sessions of psychological counseling or brief
psychotherapy. After 6 months of treatment (follow-up), the patients were
re-administered the same questionnaires.

Measures
Alexithymia was assessed with the self-report 20-item Toronto Alexithymia Scale (TAS-20) (20, 21), which is composed of 20 items rated on a
5-point Likert scale, ranging from 1 (strongly disagree) to 5 (strongly agree).
Total scores range from 20 to 100. The reliability and validity of the scale are
well established (14, 20, 21). A validated Italian version of the TAS-20 (22)
was used in this study.
Psychological distress was assessed with an Italian translation of the
Hospital Anxiety and Depression Scale (HADS) (23), which is a 14-item
self-report scale providing two separate subscale scores for anxiety
(HADS-A; 7 items) and depression (HADS-D; 7 items). The HADS is
specifically designed for assessing physically ill patients and has demonstrated reliability and validity (24, 25). The Italian translation of the HADS
was developed several years ago by the first author (P.P.) who used the
method of translation backtranslation by bilingual colleagues to establish
crosslanguage equivalence with the English version of the scale. Minor
changes were made to the wording of a few of the items after preliminary
testing of the translated scale on groups of patients and hospital staff with a
wide range of educational backgrounds. This Italian version of the HADS has
since been used for both clinical and research purposes (17, 26).
Gastrointestinal symptoms were evaluated by means of a disease-specific
observer-rated questionnaire, the Gastrointestinal Symptom Rating Scale
(GSRS) (27). The questionnaire includes 15 symptoms rated on a 7-point
Likert scale ranging from 0 (no symptoms) to 3 (pronounced symptoms),
including half points. The total score represents overall severity of gastrointestinal symptomatology, while three subscale scores correspond to dyspeptic
(epigastric and abdominal pain, squeezing sensations in the epigastrium, acid
regurgitation, heartburn, and nausea), digestive (borborygmus, abdominal
distension, eructation, and increased flatus), and intestinal (decreased passage
of stools, increased passage of stools, loose stools, hard stools, urgent need of
defecation, and feeling of incomplete evacuation) syndromes. The GSRS is
well validated and widely used in gastrointestinal research (28). For this study
an Italian translation of the GSRS was developed following a procedure
similar to that used in developing the Italian translation of the HADS. After
evaluating the translation and backtranslation of the GSRS, the first author
discussed the item content with a group of Italian gastroenterologists who
were to administer the scale. Before conducting the study, preliminary testing
of the scale was done with small groups of patients and hospital staff
representing a range of educational backgrounds.
912

Outcome Criteria
A major problem with prospectively designed studies of FGID is the lack
of consensus on how to measure outcome. However, a panel of experts (29)
has recently provided helpful guidelines including the following recommendations that are applicable to this study: (a) the main results of the study
should be based on the primary outcome measure constituted mainly by an
assessment that integrates the key symptoms of FGID and is based mainly on
patients reports of symptoms; and (b) the study protocol should include a
priori the definition of a responder, namely the change in the outcome
measure that is considered clinically meaningful. In the present study, the
GSRS total score was used to categorize patients into improved and unimproved outcome groups. Adopting the recommended guidelines, we defined
two criteria that patients had to satisfy to be included in the improved group.
The first criterion was determined by the change of overall gastrointestinal
symptoms, expressed as the proportion of change from base-line to follow-up
and calculated as follows:
(GSRS total score at base-line minus the GSRS total score at follow-up)/
GSRS total score at base-line) 100.
The mean GSRS total score change was 53% and the median score change
was 68% (range: -100% to 100%). Because a positive change of two thirds
could be reasonably considered as a good clinical improvement of symptoms,
we considered improved those patients who showed 68% (median value) of
symptom change. By itself, however, this criterion is not sufficient to define
a responder because overall symptoms at base-line may vary greatly among
patients. A second criterion was therefore chosen, which was a low level of
symptoms at follow-up. The mean GSRS total score at follow-up was 5.8
(SD 6.1) and the median was 3 (range: 0 25; higher score represents higher
level of symptoms). We considered improved those patients who obtained a
GSRS total score 3 at follow-up. Thus, using both criteria, the improved
group included patients with at least 68% of GSRS total score change from
base-line to follow-up, and a GSRS total score of 3 at follow-up.

RESULTS
Eighteen patients (4 men and 14 women) did not present
for the 6-month follow-up evaluation, two of whom had
moved from the catchment area. No significant differences
were found between the patients who did not participate in the
follow-up and those who completed the study for age, gender,
education, illness duration, types of FGID, and base-line anxiety (HADS-A), alexithymia (TAS-20), and gastrointestinal
symptoms (GSRS, both the total as well as the three subscale
scores). Those who did not complete the study had lower
base-line depression (as measured by HADS-D scores) than
those who completed the study (mean 6.7, SD 4.8 vs.
mean 10.6, SD 5.7, respectively) [t(128) 2.76, p .01].
On the basis of the two outcome criteria defined previously, the 112 patients who completed the study were divided
into improved and unimproved groups. The improved group
was comprised of 68 patients (25 men and 43 women; mean
age 37.3 years, SD 14.7; mean education 10.0 years,
SD 3.9). The unimproved group was comprised of 44
patients (15 men and 29 women; mean age 37.7 years,
SD 14.8; mean education 9.2 years, SD 3.4). Gender,
age, and education were not significantly different between
the improved and unimproved groups. In the improved group,
25 patients (36.8%) fulfilled the Rome I criteria (5) for
functional dyspepsia, 25 patients (36.8%) for irritable bowel
syndrome, 8 patients (11.7%) for both functional dyspepsia
and irritable bowel syndrome, and 10 patients (14.7%) for
functional abdominal pain. In the unimproved group, 19 patients (43.2%) fulfilled the Rome I criteria for functional
Psychosomatic Medicine 65:911918 (2003)

ALEXITHYMIA AND FUNCTIONAL GASTROINTESTINAL DISORDERS

dyspepsia, 4 patients (9.1%) for irritable bowel syndrome, 14
patients (31.8%) for both functional dyspepsia and irritable
bowel syndrome, and 7 patients (15.9%) for functional abdominal pain. The proportion of subgroups of FGID patients
who were classified as either improved or unimproved was
significantly different, 2 (3) 13.68 (p .003). Examination
of individual cell sizes for the subgroups of the FGID patients
who were improved or unimproved suggested that the difference was attributable to a higher proportion of patients with
both functional dyspepsia and irritable bowel syndrome in the
unimproved group, and a significantly higher proportion of
patients with irritable bowel syndrome alone in the improved
group. Because of the small number of subjects and the high
overlap of symptoms that is generally found across FGID
subgroups (5, 30, 31), all FGID patients were collapsed into
the two outcome-related groups.
Mean Scores and Correlations among the Measures
Table 1 displays the means and standard deviations for the
TAS-20, HADS-A, HADS-D, and GSRS total scores for
the total sample and for the improved and the unimproved
patients at base-line and at follow-up. At base-line, compared
with the improved patients, the unimproved patients displayed
significantly higher scores on the HADS-A [t (111) 1.98,
p .04], HADS-D [t (111) 4.39, p .001], TAS-20 [t (111)
7.89, p .001], and GSRS [t (111) 3.62, p .005].
These results suggest that higher levels of anxiety, depression,
alexithymia, and symptom severity may all serve as predictors
of treatment outcome. They may also suggest, however, that
these constructs are part of the illness. Based on the empirically established cut-off score of 61 on the TAS-20 (14),
56% of the total sample of FGID patients scored in the high
alexithymia range before treatment.
Table 2 shows the correlations among the measures both at
base-line and at follow-up. Although only the TAS-20 correTABLE 1.

lated significantly with the GSRS total score at base-line, the

HADS-A and HADS-D as well as the TAS-20 correlated
significantly with the GSRS total score at follow-up.
Stability
We first examined the change in mean scores from baseline to follow-up (ie, absolute stability) of the TAS-20,
HADS-A, HADS-D, and GSRS total scores for the total
sample, and for the improved and unimproved groups. These
results are displayed in Table 1. As would be expected, there
were statistically significant changes in the improved patients
for all of these variables. It is interesting to note, however, that
the magnitude of change (calculated with Cohens d) for the
TAS-20 scores was substantially smaller than it was for the
HADS-A, HADS-D, and GSRS total scores. For the unimproved patients, also as would be expected, there were no
significant changes in HADS-A, HADS-D, and GSRS total
scores and the magnitudes of these changes were substantially
smaller than they were in the improved group of patients. Only
TAS-20 scores changed significantly in the unimproved
group; however, the magnitude of change was very close to
that in the improved group, suggesting that any lack of absolute stability for TAS-20 scores is probably not associated
with change in symptom scores.
To assess relative stability of the TAS-20, we calculated
test-retest reliabilities. We correlated base-line and follow-up
scores for the entire sample; the reliability coefficient (Pearsons correlation) was r 0.76, p .001, suggesting significant relative stability. Examination of a scatterplot of correlations between base-line and follow-up TAS-20 scores
revealed that the association between base-line and follow-up
scores for those who initially scored low on the TAS-20
covaried in a similar fashion to those who initially scored high
on the TAS-20.
Another procedure that can be conducted to assess the

Means, standard deviations, t-tests, effect sizes, and test-retest correlations for the TAS-20, HADS anxiety and depression, and GSRS
total score at baseline and follow-up
Baseline

Follow-up

Scale
Mean

SD

TAS-20
HADSanxiety
HADSdepression
GSRS Total Score

58.89
10.15
10.63
11.00

13.50
4.98
5.68
5.18

TAS-20
HADSanxiety
HADSdepression
GSRS Total Score

52.19
9.41
8.91
9.65

12.10
4.50
5.20
4.15

TAS-20
HADSanxiety
HADSdepression
GSRS Total Score

69.25
11.30
13.30
13.09

7.81
5.49
5.40
5.93

Mean

3.56**
6.42**
5.59**
12.12**

0.68
1.22
1.25
2.30

2.89**
8.52**
7.37**
18.32**

0.71
2.08
1.80
4.48

2.10*
.61
1.61
3.12

0.64
0.19
0.49
0.95

SD

Total sample (n 112)

55.85
12.32
7.67
5.08
8.02
5.61
5.77
6.14
Improved Patients (n 68)
48.66
8.91
5.53
3.65
5.18
3.91
1.75
1.26
Unimproved Patients (n 44)
66.95
7.75
10.95
5.27
12.41
4.99
11.98
5.45

* p .05, **p .01; d Cohens d

Psychosomatic Medicine 65:911918 (2003)

913

P. PORCELLI et al.
and linear regression analyses were performed. For the logistic
regressions, treatment outcome (improved/unimproved)
served as the dependent (criterion) variable and base-line
TAS-20, HADS-A, HADS-D, and GSRS scores served as the
independent (predictor) variables. In the subsequent linear
regressions, percentage of change in GSRS total scores from
base-line to follow-up served as the criterion (dependent)
variable. For the logistic and linear regressions we computed
standardized scores for the TAS-20, HADS-A, HADS-D, and
GSRS.
In the first set of logistic regression analyses, we entered
each of the independent variables as single predictors to
determine how well each variable alone predicted treatment
outcome. As shown in Table 3, TAS-20 scores proved to be
the strongest predictor of treatment outcome, more than doubling the Cox and Snell R2 values and odds ratio estimates for
HADS-A, HADS-D, and GSRS total scores. As evident from
Table 2, however, the TAS-20 is significantly correlated with
these other independent variables.
Therefore, a set of hierarchical regression analyses were
performed to determine if TAS-20 base-line scores were a
unique predictor of treatment outcome, independent of the
variance shared with these other variables. HADS-A,
HADS-D, and GSRS base-line scores were entered first into
the regression model as a block (Step 1), followed by the entry
of TAS-20 base-line scores (Step 2). The results from this
analysis are displayed in Table 4. The model that included the
block containing base-line HADS-A, HADS-D, and GSRS
scores (Step 1) produced a statistically significant fit, 2(3)
28.97, p .001, Cox and Snell R2 0.23, accurately predicting 84% of the improved and 66% of the unimproved
patients (overall correct classification rate 77%). The addition of base-line TAS-20 scores (Step 2) increased significantly the overall fit of the model, 2dif (1) 32.51, p .001.
The overall fit of the model containing all four variables was
significant, 2 (4) 61.48, p .001; the addition of TAS-20
scores increased the Cox and Snell R2 to 0.42 (from 0.23), and
increased the accurate prediction of the improved patients to
85% and unimproved patients to 82%; the overall correct
classification rate improved to 83%.
In a second hierarchical logistic regression analysis, the
entry order of the variables was reversed, with base-line
TAS-20 scores entered first into the model, followed by the
block entry of base-line HADS-A, HADS-D, and GSRS
scores. This analysis, when compared with the previous re-

TABLE 2. Correlations among the measures at baseline and follow-up

TAS-20
HADS-A
HADS-D
GSRS-Total Score

TAS-20

HADS-A

HADS-D

GSRS-Total Score

.48**
.62**
.63**

.28**

.64**
.52**

.38**
.32**

.54**

.32**
.01
.10

Values above the diagonal are based on scores at baseline and values below
the diagonal represent scores at follow-up; *p .05, **p .01

relative stability of personality traits, especially in treatment

outcome studies, which typically involve substantial change in
symptom severity, is hierarchical regression. For example,
Santor et al. (18) argued that stability is demonstrated when
the personality test scores at the beginning of treatment, when
patients are acutely ill, can predict test scores on the same
measure of personality at treatment completion, when symptoms have diminished, even after controlling for symptom
severity at both the beginning and end of treatment. To this
end, we constructed regression models in which TAS-20
scores at follow-up served as the dependent (criterion) variable and base-line and follow-up HADS-D or HADS-A scores
and TAS-20 scores at base-line served as the independent
(predictor) variable. In the first set of analyses, which used the
entire sample, base-line TAS-20 scores were a significant
predictor of follow-up TAS-20 scores even after controlling
for base-line and follow-up HADS-D scores, R2chg 0.23, p
.001, and base-line and follow-up HADS-A scores, R2chg
0.35, p .001. A similar pattern of results emerged in the
improved group, when controlling for base-line and follow-up
HADS-D scores, R2chg 0.27, p .001, and base-line and
follow-up HADS-A scores, R2chg 0.32, p .001; and also in
the unimproved group, when controlling for base-line and
follow-up HADS-D scores, R2chg 0.22, p .001, and baseline and follow-up HADS-A scores, R2chg 0.28, p .001.
These results suggest that any association between anxiety and
depression scores and TAS-20 scores cannot account for the
stability of TAS-20 scores across the treatment interval. Thus,
TAS-20 scores can be seen as a reliable and stable variable for
predicting treatment outcome.
Predicting Treatment Outcome
Having established the relative stability of alexithymia, as
measured by the TAS-20, we next sought to determine if
alexithymia can predict treatment outcome. A series of logistic
TABLE 3.

Facet/Domain

HADSAnxiety
HADSDepression
GSRSTotal
TAS-20

Results from individual variable logistic regressions predicting treatment outcome

R2

.03
.14
.10
.38

Final Model

2
3.93*
16.84**
12.21**
53.64**

df

SE

Wald

OR

1
1
1
1

.39
.86
.72
2.09

.20
.23
.22
.41

3.76
14.52**
10.72**
25.79**

1.48
2.36
2.05
8.05

R2 Cox & Snell R2; *p .05, **p .01

914

Psychosomatic Medicine 65:911918 (2003)

ALEXITHYMIA AND FUNCTIONAL GASTROINTESTINAL DISORDERS

TABLE 4.

Step

Step 1

Step 2

Facet/Domain

Block
HADSanxiety
HADSdepression
GSRStotal
TAS-20

Results from hierarchical logistic regressions predicting treatment outcome

R2

.23

.42

df

28.97**

R2chg

2chg

Final Model
df
B

SE

Wald

OR

.02
.60
.52
1.88

.30
.30
.28
.43

.00
4.05*
3.51
19.42**

.98
1.82
1.69
6.55

61.48**

.19

32.50**

R2 Cox & Snell R2; *p .05, **p .01

gression equation, allows one to gauge the relative predictive

capacity of the block of HADS-A, HADS-D, and GSRS
scores with TAS-20 scores in the prediction of treatment
outcome. The model that included only the TAS-20 scores
(Step 1) produced a statistically significant fit, 2 (1) 53.64,
p .001, Cox and Snell R2 0.38, accurately predicting 74%
of the improved and 89% of the unimproved patients (overall
correct classification rate 80%). In Step 2, the addition of
the block containing HADS-A, HADS-D, and GSRS increased the overall fit of the model, 2dif (1) 7.84, p .05.
It is important to note, however, that the R2 value for alexithymia as a single predictor was 0.38 compared with an R2
value of 0.23 for the symptom variables measured by the
HADS-A, HADS-D, and GSRS. Thus, whereas the addition of
TAS-20 scores increased the R2 value from 0.23 to 0.42 when
added to the block entry of HADS-A, HADS-D, and GSRS,
adding these latter variables only increased the R2 values from
0.38 to 0.42.
The above analyses examined treatment outcome categorically with patients classified as either improved or unimproved. Since patients with more severe FGID symptoms were
also more likely to be psychologically distressed, there was a
potential to increase artifactually the association between the
psychological measures and categorically defined outcome. In
addition, given that one criterion of a treatment responder was
that the GSRS total score be no greater than 3, it is possible
that patients with milder gastrointestinal symptoms at the
beginning of treatment were more likely to be categorized as
improved at the end of treatment. We therefore performed a
series of linear multiple regression analyses using percent
change in symptoms as the outcome measure.
Using the same modeling procedure as we used with the
logistic regression analyses, we entered the HADS-A,
HADS-D, and TAS-20 base-line scores as single predictors
TABLE 5.

to determine how well each of these independent variables

alone predicted treatment outcome. As percentage change
in GSRS total scores from base-line to follow-up was used
exclusively to construct the criterion variable (ie, dependent variable), GSRS scores could not be used as a predictor variable. As shown in Table 5, the results are similar to
those obtained using the categorical outcome measure in
the logistic regression, with TAS-20 scores emerging as the
best predictor of treatment outcome compared with the HADS-A
and HADS-D. We next performed a series of hierarchical
regression analyses to determine if TAS-20 scores were a
unique predictor of percentage change in GSRS scores from
base-line to follow-up. HADS-A and HADS-D base-line
scores were entered first into the regression model as a
block (Step 1), followed by the entry of TAS-20 base-line
scores (Step 2). The results from this analysis are displayed
in Table 6. Again, even after controlling for base-line
HADS-A and HADS-D scores, base-line TAS-20 score
proved to be a significant predictor, adding a significant
increase in incremental variance in the prediction of percentage change in GSRS scores.
DISCUSSION
The results of this study indicate that FGID patients who
did not respond to treatment were more alexithymic, more
depressed, and had more pronounced gastrointestinal symptoms both before and after a period of 6 months of treatment
than FGID patients who did respond to treatment. Although
the findings suggested that all of these variables might serve
as base-line predictors of treatment outcome, alexithymia
clearly emerged as the most powerful predictor of both recovery status and overall reduction in GSRS symptoms at followup. More impressively, alexithymia, assessed at base-line, was
able to predict recovery status and reduction in symptoms

Results from individual linear regressions predicting percentage change in GSRS total scores from baseline to follow-up
Final Model

Facet/Domain

HADSanxiety
HADSdepression
TAS-20

R2

.04
.17
.29

4.93*
22.14**
45.36

df

1,110
1,110
1,110

SE

1.74
3.01
1.67

.78
.64
.25

.21
.41
.54

* p .05, **p .01

Psychosomatic Medicine 65:911918 (2003)

915

P. PORCELLI et al.
TABLE 6.

Results from hierarchical linear regressions predicting percentage change in GSRS total scores from baseline to follow-up
Final Model

Step

Step 1

Step 2

Facet/Domain

Block
HADSanxiety
HADSdepression
TAS-20

R2

.17

.34

11.47**

18.55**

df

R2chg

df
B

SE

.07
1.73
.14

.70
.64
.27

.01
.24
.45

2,109

3,108

.17

27.19**

1,108

* p .05, **p .01

even after controlling for base-line gastrointestinal symptoms,

depression, and anxiety symptoms. Moreover, evidence that
alexithymia can be considered a reliable and stable predictor
of treatment outcome was provided by significant correlations
between base-line and follow-up TAS-20 scores, and by the
results of hierarchical regression analyses showing that the
stability of TAS-20 scores over the 6-month treatment period
could not be accounted for by their associations with anxiety
and depression scores. The main contribution of TAS-20
scores was to improve the prediction of patients who did not
respond to treatment compared with patients who responded
to treatment.
The finding that 56% of the FGID patients scored in the
high range of the TAS-20 is lower than the finding of 66% in
a previous study of FGID patients (17), but this rate of high
alexithymia is similar to rates that have been reported in
studies of patients with panic disorder, posttraumatic stress
disorder, essential hypertension, somatoform pain disorder,
substance use disorders, and eating disorders (for reviews see
14,15). Despite the strong associations between alexithymia
and these various disorders, however, only a few studies have
examined the role of alexithymia in predicting treatment outcome. Kosten et al. (32) studied a group of Vietnam combat
veterans with posttraumatic stress disorder who entered a
double-blind, placebo-controlled pharmacological trial of 8
weeks. Alexithymia was measured with the Alexithymia Provoked Response Questionnaire (33) and was found to be
stable over the 8 week treatment period and to significantly
predict treatment outcome. In another study, Bach and Bach
(34) investigated a group of psychiatric inpatients who had
presented with functional somatic symptoms and received
diagnoses of anxiety or somatoform disorders. The patients
were treated with cognitive and behavioral therapies for a
minimum of 8 weeks and reassessed 2 years later. Alexithymia, measured by the 26-item version of the Toronto Alexithymia Scale (35), was found to be a significant predictor of
persistent somatization and was independent of sociodemographic variables, other psychopathology, and illness severity.
There are several potential pathways by which alexithymia
might influence symptom severity and treatment outcome for
FGID patients. These include a limited ability of high alexithymia individuals to cope adaptively with stressful situations
(36, 37), which may contribute to high levels of psychological
distress and a possible sustained arousal of the physiological
component of emotion response systems (15, 38). In addition,
916

high alexithymia individuals may be prone to functional somatic symptoms because of a tendency to amplify, focus on,
and misinterpret the somatic sensations that accompany states
of emotional arousal as well as other normal bodily sensations
(38, 39). There is evidence that FGID patients have a low pain
threshold in the perception of visceral stimuli, including in
bodily sites beyond the symptom-related region of the gastrointestinal tract (40). Furthermore, the bidirectional brain-gut
axis allows somatic symptoms to be generated not only by
motility disturbances and alterations in sensory inputs from
the gut, but also by unregulated states of emotional arousal
and by mutual interactions between these and other behavioral
factors, such as abnormal illness behavior and health careseeking behavior (41, 42). Consequently, patients with high
alexithymia may experience more severe somatic symptoms
and respond poorly to treatment because of the difficulty in
cognitively processing emotional and somatic stimuli.
Another pathway whereby alexithymia might contribute to
a persistence of FGID symptoms is through an association
with unhealthy behaviors such as poor nutritional consumption (eg, a high-fat diet or eating foods high in sugar content),
poor eating behavior (eg, fast eating and bingeing), alcohol
and drug use, and a sedentary lifestyle (43). For example,
Tang et al. (44) found that several psychological, behavioral,
and cognitive aspects of eating disorders were present among
patients with irritable bowel syndrome. Crowell et al. (45)
studied FGID in obese binge eaters, obese nonbinge eaters,
and nonobese binge eaters and found that functional symptoms of the digestive tract were dependent on binge eating
behavior and not on body weight. Porcelli et al. (46) found
that lifetime eating disorders were significantly more prevalent in patients with FGID than in patients with gallstones.
Moreover, as noted earlier, alexithymia is associated strongly
with both eating disorders and alcohol abuse; these disorders
have been conceptualized as maladaptive efforts to self-regulate distressing affects (14). Future studies might evaluate the
influence of eating disorders and alcohol abuse on treatment
outcome in patients with FGID.
In the present study, treatment outcome was also predicted
by depression. This finding is consistent with previous studies
showing that higher psychological distress and psychiatric
comorbidity may be associated with persistent and severe
FGID symptoms (42, 47). Moreover, whereas both the improved and unimproved groups of patients showed only modest reductions in alexithymia, the improved patients had much
Psychosomatic Medicine 65:911918 (2003)

ALEXITHYMIA AND FUNCTIONAL GASTROINTESTINAL DISORDERS

larger reductions in depression and anxiety than the unimproved patients. It is possible that a reduction in depression
and anxiety results in fewer FGID symptoms, or that an
improvement in FGID relieves associated depression and
anxiety.
There are several limitations to this study that should
caution against generalizing from the results. First, the FGID
patients were recruited from a tertiary care clinic and may
therefore represent the most severe groups of FGID patients in
which a high level of alexithymia may be more prevalent.
FGID patients referred to tertiary care clinics have been found
to have more psychological distress, psychiatric disorders, and
abnormal illness behavior than FGID patients referred to primary and secondary care settings (5, 48). In our particular
sample, however, the severity of gastrointestinal symptoms
and level of psychological distress were not very high, and
60% responded to treatment, although no comparative data are
available on the epidemiology of FGID in the region where
the study was performed. There is some evidence also that
alexithymia may be linked more with increased illness behavior than with severity of illness. Lumley and Norman (49), for
example, in a study of college students, found that the alexithymic difficulty in identifying feelings was associated with
increased use of outpatient treatment, even after controlling
for depression and somatic complaints.
A second limitation is that because the study was not
intended as a treatment controlled investigation, patients underwent different forms of treatment (including gastrointestinal and psychotropic medications, psychological counseling,
and brief psychotherapy) that were combined in various ways
on a case by case basis. Moreover, during the 6-month treatment period, the combinations of therapies sometimes
changed for individual patients, in accordance with changes in
their symptomatology and overall health. This therapeutic
approach to FGID is quite usual in clinical practice, and the
study therefore could not be regarded as a controlled investigation of one specific therapy.
Third, the methodology of the study was limited by a
failure to control for abnormal eating behavior, alcohol use,
coffee consumption, and smoking. Since these behaviors may
aggravate gastrointestinal symptoms, they should be controlled for in future studies.
Fourth, the interval of 6 months between base-line and
follow-up assessments was quite short and precludes generalizing to the prediction of treatment outcomes over longer
periods. In addition, follow-up outcomes were assessed with a
single observation and may not be reliable because FGID
symptoms tend to fluctuate. Further studies with longer follow-up intervals and multiple outcome assessments are
needed to establish more conclusively that alexithymia is
stable in the context of symptom change.
Finally, the small numbers of patients in some of the FGID
subgroups precluded examining whether there were any statistically significant differences in outcome for each of the
subgroups. Nonetheless, the higher number of patients with
both functional dyspepsia and irritable bowel in the unimPsychosomatic Medicine 65:911918 (2003)

proved group raises the possibility that treatment outcome

might be influenced by the number of coexisting FGID subgroups. In the present study, however, the severity of gastrointestinal symptoms was not a significant predictor of treatment outcome.
In conclusion, the findings from the present study suggest
that alexithymia is a stable personality trait in FGID patients,
and like depression, it may be a useful predictor of treatment
outcome. However, because the patients in this study were
recruited from a tertiary care hospital and the medical and
psychological treatments were tailored to each patients needs,
further studies should evaluate the association between alexithymia and treatment outcome in FGID patients from different
medical settings and in randomized, controlled treatment trials. Nonetheless, because no treatment has been shown to be
definitely effective for FGID, and multicomponent therapeutic
strategies are usually employed, the present findings may have
important clinical implications. In addition to treating comorbid anxiety and depression, clinicians might improve treatment outcome for FGID patients by identifying those patients
who are high in alexithymia, and attempting to improve these
patients skills for coping with emotionally stressful
situations.
REFERENCES
1. Drossman DA, Li Z, Andruzzi E, Temple RD, Talley NJ, Thompson WG,
Whitehead WE, Janssens J, Funch-Jensen P, Corazziari E, Richter JE,
Koch GG. U.S. householder survey of functional gastrointestinal
disorders: prevalence, sociodemography and health impact. Dig Dis Sci
1993;38:1569 80.
2. Talley NJ, Zinstmeister AR, Schleck CD, Melton III LJ. Dyspepsia and
dyspepsia subgroups: a population-based study. Gastroenterology 1992;
102:1259 68.
3. Mitchell CM, Drossman DA. Survey of the AGA membership relating to
patients with functional gastrointestinal disorders. Gastroenterology
1987;92:1282 84.
4. Thompson WG, Heaton KW, Smyth GT, Smyth C. Irritable bowel
syndrome in general practice: prevalence, characteristics, and referral.
Gut 2000;46:78 82.
5. Drossman DA, ed. The Functional Gastrointestinal Disorders: Diagnosis, Pathophysiology, and Treatment. Boston: Little, Brown and
Company;1994.
6. Ringel Y, Drossman DA. From gut to brain and back. A new perspective
into functional gastrointestinal disorders. J Psychosom Res 1999;47:
20510.
7. Kellow JE, Delvaux M, Azpiroz F, Camilleri M, Quigley EMM, Thompson DG. Principles of applied neurogastroenterology: physiology/
motility-sensation. Gut 1999;45(Suppl II):II17II24.
8. Talley NJ, Spiller R. Irritable bowel syndrome: a little understood organic
bowel disease? Lancet 2002;360:555 64.
9. Drossman DA, Creed FH, Olden KW, Svedlund J, Toner BB, Whitehead
WE. Psychosocial aspects of the functional gastrointestinal disorders. Gut
1999;45(Suppl. II):II25II30.
10. Osterberg E, Blomquist L, Krakau I, Weinryb RM, Asberg M, Hultcrantz
R. A population study on irritable bowel syndrome and mental health.
Scand J Gastroenterol 2000;35:264 68.
11. Wood JD, Alpers DH, Andrews PLR. Fundamentals of neurogastroenterology. Gut 1999;45(Suppl II):II6 II16.
12. Whitehead WE, Bosmajian L, Zonderman AB, Costa PT, Schuster MM.
Symptoms of psychologic distress associated with irritable bowel syndrome. Gastroenterology 1988;95:709 14.
13. Bennett EJ, Palmer K, Badcock C-A, Tennant CC, Kellow JE. Functional
gastrointestinal disorders: psychological, social, and somatic features.
Gut 1998;42:414 20.
14. Taylor GJ, Bagby RM, Parker JDA. Disorders of Affect Regulation.
917

P. PORCELLI et al.

15.
16.
17.
18.
19.
20.
21.
22.

23.
24.
25.
26.
27.
28.
29.
30.
31.

918

Alexithymia in Medical and Psychiatric Illness. Cambridge: Cambridge

University Press;1997.
Taylor GJ. Recent developments in alexithymia theory and research. Can
J Psychiatry 2000;45:134 42.
Nemiah JC, Freyberger H, Sifneos PE. Alexithymia: a view of the
psychosomatic process. In: Hill OW, ed. Modern Trends in Psychosomatic Medicine. Vol 3. London: Butterworths;1976:430 39.
Porcelli P, Taylor GJ, Bagby RM, De Carne M. Alexithymia and functional gastrointestinal disorders. A comparison with inflammatory bowel
disease. Psychother Psychosom 1999;68:263 69.
Santor DA, Bagby RM, Joffe RT. Evaluating stability and change in
personality and depression. J Pers Soc Psychol 1997;73:1354 62.
Luminet O, Bagby RM, Taylor GJ. An evaluation of the absolute and
relative stability of alexithymia in patients with major depression. Psychother Psychosom 2001;70:254 60.
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;38:2332.
Bagby RM, Parker JDA, Taylor GJ. The twenty-item Toronto Alexithymia Scale. II. Convergent, discriminant, and concurrent validity. J Psychosom Res 1994;38:33 40.
Bressi C, Taylor GJ, Parker JDA, Bressi G, Brambilla V, Aguglia E,
Allegranti I, Bongiorno A, Giberti F, Bucca M, Todarello O, Callegari C,
Vender S, Gala C, Invernizzi G. Cross validation of the factor structure
of the 20-item Toronto Alexithymia Scale: An Italian multicenter study.
J Psychosom Res 1996;41:55159.
Zigmond AS, Snaith RP. The Hospital Anxiety and Depression Scale.
Acta Psychiatr Scand 1983;67:36170.
Hermann C. International experiences with the Hospital Anxiety and
Depression Scale. A review of validation data and clinical results. J Psychosom Res 1997;42:17 41.
Naifeh KH. Psychometric testing in functional GI disorders. In: Olden
KW, ed. Handbook of Functional Gastrointestinal Disorders. New York:
Marcel Dekker;1996:79 126.
Porcelli P, Leoci C, Guerra V, Taylor GJ, Bagby RM. A longitudinal
study of alexithymia and psychological distress in inflammatory bowel
disease. J Psychosom Res 1996;41:569 73.
Svedlund J, Sjdin I, Dotevall G. GSRS. A clinical rating scale for
gastrointestinal symptoms in patients with irritable bowel syndrome and
peptic ulcer disease. Dig Dis Sci 1988;33:129 34.
Wiklund I. Aspects of quality of life in gastrointestinal disease: some
methodological issues. Scand J Gastroenterol 1993;30 Suppl 208:
129 32.
Veldhuyzen van Zanten SJO, Talley NJ, Bytzer P, Klein KB, Whorwell
PJ, Zinsmeister AR. Design of treatment trials for functional gastrointestinal disorders. Gut 1999;45(Suppl II):69 77.
Agreus L, Svardsudd K, Nyren O, Tibblin G. Irritable bowel syndrome
and dyspepsia in the general population: overlap and lack of stability over
time. Gastroenterology 1995;109:671 80.
Holtmann G, Goebell H, Talley NJ. Functional dyspepsia and irritable

32.

33.

34.
35.
36.
37.
38.

39.

40.

41.

42.

43.
44.

45.

46.
47.
48.

49.

bowel syndrome: is there a common pathophysiological basis? Am J

Gastroenterol 1997;92:954 59.
Kosten TR, Krystal JH, Giller EL, Dan E. Alexithymia as a predictor of
treatment response in post-traumatic stress disorder. J Traum Stress
1992;5:4150.
Krystal JH, Giller EL, Cicchetti DV. Assessment of alexithymia in
post-traumatic stress disorder and somatic illness: introduction of a reliable measure. Psychosom Med 1986;48:84 94.
Bach M, Bach D. Predictive value of alexithymia: a prospective study in
somatizing patients. Psychother Psychosom 1995;64:43 8.
Taylor GJ, Ryan DP, Bagby RM. Toward the development of a new
self-report alexithymia scale. Psychother Psychosom 1985;44:19199.
Parker JDA, Taylor GJ, Bagby RM. Alexithymia: relationship with ego
defense and coping styles. Compr Psychiatry 1998;39:91 8.
Parker JDA, Taylor GJ, Bagby RM. The relationship between emotional
intelligence and alexithymia. Pers Indiv Diff 2001;30:10715.
Lumley MA, Stettner L, Wehmer F. How are alexithymia and physical
illness linked? A review and critique of pathways. J Psychosom Res
1996;41:50518.
Nakao M, Barsky AJ, Kumano H, Kuboki T. Relationship between
somatosensory amplification and alexithymia in a Japanese psychosomatic clinic. Psychosomatics 2002;43:55 60.
Trimble KC, Farouk R, Pryde A, Douglas S, Heading RC. Heightened
visceral sensation in functional gastrointestinal disease is nonsitespecific. Evidence for a generalized disorder of gut sensitivity. Dig Dis
Sci 1995;40:160713.
Herschbach P, Henrich G, von Rad M. Psychological factors in functional
gastrointestinal disorders: characteristics of the disorder or of the illness
behavior? Psychosom Med 1999;61:148 53.
Drossman DA, Whitehead WE, Toner BB, Hu YJ, Bangdiwala SI, Jia H.
What determines severity among patients with painful functional bowel
disorders? Am J Gastroenterol 2000;95:974 80.
Helmers KF, Mente A. Alexithymia and health behaviors in healthy male
volunteers. J Psychosom Res 1999;47:635 45.
Tang TN, Toner BB, Stuckless N, Dion KL, Kaplan AS, Ali A. Features
of eating disorders in patients with irritable bowel syndrome. J Psychosom Res 1998;45:17178.
Crowell MD, Cheskin LJ, Musial F. Prevalence of gastrointestinal symptoms in obese and normal weight binge eaters. Am J Gastroenterol
1994;89:38791.
Porcelli P, Leandro G, De Carne M. Functional gastrointestinal disorders
and eating disorders. Scand J Gastroenterol 1998;33:577 82.
Camilleri M. Management of the irritable bowel syndrome. Gastroenterology 2001;120:652 68.
Drossman DA, Thompson WG. Irritable bowel syndrome: a graduated,
multi-component treatment approach. Ann Intern Med 1992;116:
1009 16.
Lumley MA, Norman S. Alexithymia and health care utilization. Psychosom Med 1996;58:197202.

Psychosomatic Medicine 65:911918 (2003)