Professional Documents
Culture Documents
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-
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)
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
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-
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
Facet/Domain
HADSAnxiety
HADSDepression
GSRSTotal
TAS-20
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
Step
Step 1
Step 2
Facet/Domain
Block
HADSanxiety
HADSdepression
GSRStotal
TAS-20
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**
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
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
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)
P. PORCELLI et al.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
918
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.