s i
bid
Em
arbu
3; acc
ession
The way people think about illness and how they Current concepts of somatoform disorders emphasize
Journal of Psychosomatic Researchated disability and coping behavior. In particular, this seems
to be the case for the attribution of common somatic
symptoms, such as abdominal pain, back pain, chest pain,
or headache. These and other unexplained physical symp-
toms are key features of somatoform disorders in DSM-IV.
Somatoform symptoms are common phenomena with base
rates of above 30% for single symptoms [1]; the symptoms
to use organically oriented causal attributions for common
somatic complaints. However, empirical approaches to
evaluate this assumption are rare.
Robbins and Kirmayer [4] investigated attributions of
common somatic symptoms using the symptom interpreta-
tion questionnaire. They revealed three dimensions of causal
attributions, namely, psychological, somatic, and normaliz-a major determinant of health-care-seeking, illness-associ-
interpret illness-associated somatic symptoms seems to be the role of causal illness attributions [2,3]. It is assumed
that patients with somatoform symptoms have a tendencyIntroduction are associated with substantial disability and significant
costs for the health care system.attending treatment in primary care. Inclusion criteria were
‘‘unexplained physical symptoms.’’ All patients were investigated
using structured interviews and self-rating scales [Screening for
Somatoform Symptoms (SOMS), Beck Depression Inventory
(BDI), Beck Anxiety Inventory, and a 12-item instrument to assess
causal attributions]. By means of factor analysis, the following
illness attributions were considered: vulnerability to infection and
environmental factors, psychological factors, organic causes
including genetic and aging factors, and distress (including
exhaustion and time pressure). Results: Most patients reported
Keywords: Illness attribution; Somatoform disorders; Illness behavior; Depr0022-3999/04/$ – see front matter D 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.jpsychores.2004.02.015
* Corresponding author. Department of Clinical Psychology and
Psychotherapy, Philipps-University of Marburg, Gutenbergstr. 18, 35032
Marburg, Germany.
E-mail address: rief@mailer.uni-marburg.de (W. Rief ).depression and with anxiety disorders was associated with more
psychological attributions. Evenwhen the influence of somatization,
depression, and anxiety is controlled for, illness beliefs still showed
associations with illness behavior. Organic causal beliefs and
vulnerability attributions were associated with a need for medical
diagnostic examinations, increased expression of symptoms,
increased illness consequences, and bodily scanning. Conclusions:
Multiple causal attributions can coexist demonstrating different
associations with comorbid depression and illness behavior.
D 2004 Elsevier Inc. All rights reserved.
; Primary careAbstract
Objective: To compare causal illness beliefs between patients
with unexplained physical symptoms and different comorbid
disorders and to assess the association of causal illness beliefs with
illness behavior. Methods: We examined a sample of 233 patients
multiple illness attributions. The more somatoform symptoms
patients had, the more explanations in general they considered.
Especially for vulnerability and organic illness beliefs, patients with
somatoform symptoms had increased scores. Comorbidity withCausal illness attribution
Associations with comor
Winfried Rief*, Alexandra Nanke, Julia
Philipps-University of M
Received 4 August 200n somatoform disorders
ity and illness behavior
merich, Andrea Bender, Thomas Zech
rg, Marburg, Germany
epted 3 February 2004
57 (2004) 367–371ing attributions. Sensky et al. [5] could demonstrate that
causal attributions about common somatic sensations are
associated with the frequency of general practice visits. The
most striking difference between high utilizers and low
utilizers of the health care system was found on the
dimension of normalizing attributions: Frequent attenders
reported less normalizing explanations for common bodily
sensations than the comparison group.

ations for one and the same symptom. Therefore, we
somawanted to examine the multiplicity of causal illness
attributions in somatoform disorders. We expected the
following results:

The more symptoms patients have, the more illness
attributions they consider.

Moreover, we expect that illness attributions are
associated with comorbidity patterns: Patients with
pure somatoform disorders are expected to show more
organically oriented explanations for symptoms, while
patients with somatoform and comorbid depressive
disorders are expected to show more psychological
explanations for the symptoms.

Finally, we also wanted to address possible associa-
tions between illness attribution and several aspects of
illness behavior. Therefore, it was necessary to assess
illness behavior not as unidimensional, but as a
multidimensional construct with aspects such as
seeking diagnostic or treatment options, illness
consequences, and others (see Ref. [8])
Methods
Subjects, design, and procedure
The first step was the acquisition of collaborating GP
offices. From more than 200 contacted GPs, 25 accepted to
participate in our study. GPs were instructed to approach
all patients with at least two unexplained physical symp-
toms. Two hundred and ninety five patients (65% female;
mean age 50.6 years; S.D.=15.4; range 17–82 years)
fulfilled this criterion. The number of recruited patients
per doctor ranged from 3 to 15. Patients with somatic
disorders difficult to differentiate from somatoform disor-
ders were not included. Most patients were married or
living with a partner (66%) and about 45% had higher
education. If patients agreed to participate, they wereThe cognitive representation of illness is typically con-
ceptualized following Leventhal’s self-regulatory model.
Following his model, the Illness Perception Questionnaire
(IPQ) was developed, which covers five dimensions: (1)
‘‘identity’’ of the illness comprises 12 symptom items; (2)
the dimension ‘‘time line’’ includes items asking for the
expected course (persisting, temporary, short-time); other
dimensions are (3) ‘‘cause’’ of the symptoms (organic,
psychological), (4) expected consequences of the illness,
and (5) expected cure/control. These dimensions of the
representation of illness seem to be stable features [6] and
can predict other aspects of illness course and illness
behavior [7].
To date, illness attribution is typically assessed two or
three dimensionally. However, symptom attribution seems
to be a multidimensional process with coexisting explan-
W. Rief et al. / Journal of Psycho368informed about the study, interviewed, and got the self-rating scales (see below) either immediately in the GP
office, at home, or in the university department. The final
sample not only fulfilling inclusion criterion but also
answering the self-rating scales was 233 patients (same
age and sex distribution as sample above). During the
following 6 months, the GPs were again interviewed to
check whether the physical symptoms were still ‘‘organi-
cally unexplained.’’ It is intended to analyze longitudinal
data; however, in this manuscript we focus on cross-
sectional data.
Assessment instruments
All patients were interviewed using a standardized
interview instrument (International Diagnostic Check List,
IDCL [9]) to get reliable and valid DSM-IV diagnoses.
This interview guideline checks the criteria of the 30 most
frequent mental disorders. Its reliability and validity is
well established.
The Screening for Somatoform Symptoms (SOMS) [10]
is a self-rating scale assessing 53 organically unexplained
physical symptoms. In this study, we used the state version
of the SOMS where subjects rated the intensity of the single
symptom during the past 7 days on a five-point Likert scale.
This scale has been validated not only for status diagnosis,
but also for the assessment of change [11]. The SOMS state
version results in two dependent variables, namely, the
symptom count (sum of all positively answered symptoms)
and the somatization severity index (means of all item
answers). Reliability and validity indices of SOMS are
satisfactory [10].
Illness behavior was assessed using the Scale for the
Assessment of Illness Behavior (SAIB) [8], a 26-item scale
assessing five factors of illness behavior (diagnosis verifi-
cation, expression of complaints, medication and treatment,
consequences of illness, and body scanning).
For the assessment of illness attributions, we modified a
subscale of the IPQ [12] resulting in a 12-item scale (see
Table 1) with a five-point Likert response format; the
answers are scored from 0 to 4. This answer format allows
to disagree, to agree, or neither-agree-nor-disagree answers
to specific beliefs about the aetiology of the symptoms.
Further instruments were the Beck Depression Inventory
(BDI) [13] and the Beck Anxiety Inventory [14].
Statistic methods
The first step was the definition of factors for causal
illness attributions. Therefore, we planned to use principal
component analysis with subsequent Varimax rotation.
Afterwards, groups with different diagnoses will be de-
fined and compared for causal illness beliefs; to control
for possible age and sex differences, analysis of variance
with age and sex as covariates will be used. To analyze
associations between illness attributions and illness be-
tic Research 57 (2004) 367–371havior, partial correlations will be computed eliminating