in
e u
nte
o
Keywords: Health care utilization; Medically unexplained physical symptoms; Psychological intervention; Somatization; Symptom perception
earchIntroduction
Medically unexplained physical symptoms, i.e., symp-
toms not conclusively explained by an organic disease, are
widespread. They are considered a burden to patient and, in
case of medical consultation, health care system. Prevalence
rates range from 4% to 20% depending on the population
studied (e.g., Refs. [1–3]) and on the method of symptom
plained symptoms and the costs that go with it is emphasized
again and again (e.g., Refs. [8–12]).
Psychological intervention
Somatoform disorders have been shown to respond to
psychological intervention [13–15] and to treatment with
antidepressants [16]. However, most studies evaluating theconsultations over a period of 1 1/2 years. Results: ANOVAs for
repeated measures showed that self-reported and GP-registered
unexplained physical symptoms decreased from pretest to posttest
perception variables, whereas the prediction of change in registered
symptoms and consultations requires a different model.
D 2004 Elsevier Inc. All rights reserved.Abstract
Objectives: To test the effect of psychological intervention on
multiple medically unexplained physical symptoms, psychological
symptoms, and health care utilization in addition to medical care as
usual. To identify patient-related predictors of change in symptoms
and care utilization. Methods: In a randomized controlled trial,
subjects were assigned to one of two conditions: psychological
intervention by a qualified therapist plus care as usual by a general
practitioner (GP) or care as usual only. Participants (N= 98) were
administered a standardized interview and several outcome
measures at intake and after 6 months and 12 months after intake.
GPs rated medically unexplained and explained symptoms anddetection [4]. Unexplained symptoms are strongly and
consistently associated with emotional distress, in particular,
symptoms of depression and anxiety (e.g., Refs. [5–8]). The
increased use of medical services by patients with unex-
0022-3999/04/$ – see front matter D 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.jpsychores.2004.02.012
* Corresponding author. Tel.: +31-20-525-6812; fax: +31-20-639-1369.
E-mail address: a.m.m.kolk@uva.nl (A.M.M. Kolk).to follow-up. Psychological symptoms and consultations decreased
from pretest to posttest. GP-registered explained symptoms did not
decrease. However, intervention and control groups did not differ
in symptom reduction. Path analysis revealed two paths to a
decrease in self-reported unexplained physical symptoms: from
more negative affectivity via more psychological attribution and
more pretreatment anxiety, and from more somatic attribution via
more psychological attribution and more pretreatment anxiety.
Conclusion: Intervention and control groups did not differ in
symptom reduction. Reduction of self-reported medically unex-
plained symptoms was well predicted by patient-related symptomA.M.M. Kolk*, S. Schagen, G.J.F.P. Hanewald
Department of Clinical Psychology, University of Amsterdam, Roetersstraat 15, 1018 WB Amsterdam, The Netherlands
Received 17 February 2003; accepted 3 February 2004Multiple medically unexpla
health car
Outcome of psychological i
predictors
Journal of Psychosomatic Resed physical symptoms and
tilization
rvention and patient-related
f change
57 (2004) 379–389effect of psychological intervention pertain to symptom
syndromes and hypochondriasis. With regard to multiple
medically unexplained symptoms, only six randomized
controlled clinical trials—mostly pertaining to symptoms
lasting several years—could be identified, evaluating cog-
nitive–behavioural methods (6–16 sessions) [17–20] and
emotional disclosure (three sessions) [21,22]. These trials
considered a variety of response variables the main

chosooutcome measures being multiple medically unexplained
symptoms, psychological symptoms, and health care visits.
As assessed with self-report symptom questionnaires, a
larger reduction in physical symptoms in the intervention
groups was demonstrated in three studies [17,18,20],
whereas two studies [17,20] showed a larger reduction
in psychological symptoms. As assessed with medical
chart or self-report questionnaires, a decrease in health
care utilization (visits) was found in three studies [17,
20,21]. One study [21] assessed costs associated with
medical services and found a decrease in the treatment
group twice as great as in the control group. These mixed
results may be attributed to differences not only in
methodology and intervention-related variables but also
in patient characteristics such as age, gender, socioeco-
nomic status (SES), and personality traits.
The f irst objective of this study was to investigate the
effect of psychological interventions on several outcome
variables in subjects presenting with one or more medi-
cally unexplained symptoms of relatively recent onset (3–
12 months). A nonmanualized treatment, reflecting usual
as opposed to manualized treatment, was chosen to
enhance the external validity of the study. In contrast to
the literature, this study focused on subjects with symp-
toms of relatively recent onset. These symptoms were
expected to be more susceptible to treatment and their
study was considered relevant to the prevention of somatic
fixation. In line with the literature, the primary outcome
measure was self-reported medically unexplained physical
symptoms as assessed with a self-report symptom ques-
tionnaire [17,18,20]. However, other outcomes were stud-
ied as well, i.e., self-reported psychological symptoms,
general practitioner (GP)-registered medically unexplained
physical symptoms, GP consultations, and registered
explained physical symptoms (symptoms explained by an
acute or chronic organic disease). Since unexplained
physical symptoms are strongly and consistently associat-
ed with emotional distress, in particular symptoms of
depression and anxiety, it has been repeatedly recommen-
ded that the impact of treatment on psychological symp-
toms is more adequately addressed [13–15]. As the
prevalence of symptoms depends on the method of symp-
tom detection, the number of self-reported symptoms (for
example, as assessed with self-report questionnaires) will
not generally coincide with the number of symptoms
presented to and registered by a GP [4,23]. Therefore, it
is of considerable theoretical and practical interest to
verify that treatment is followed by a decrease in GP-
registered symptoms and GP consultations as well
[4,24,25].
In a randomized controlled trial, subjects presenting with
medically unexplained symptoms of recent onset were
assigned to one of two conditions: psychological intervention
by a qualified therapist plus care as usual by their GP, or care
as usual only. A 2 (condition: intervention and control)
3
A.M.M. Kolk et al. / Journal of Psy380(time: pretest, posttest, follow-up) design was employed.Hypotheses
It was expected that self-reported unexplained physical
symptoms in the intervention group would significantly
decrease from pretest to posttest to follow-up, whereas a
significant symptom reduction would be absent in the
control group. The same results were expected with regard
to self-reported psychological symptoms, GP-registered
medically unexplained physical symptoms, and GP consul-
tations. In addition, the possibility that the intervention
affected the GP-registered level of medically explained
physical symptoms was explored.
Patient-related predictors of change
As patients with multiple medically unexplained physical
symptoms form a heterogeneous group, it is essential to
identify patient-related predictors of change in symptoms
and care utilization in addition to the evaluation of the
outcome of interventions [24]. However, hardly any study
addressed this issue. Pretreatment anxiety has been found to
predict better outcome [26]. Other possible predictors of
change may concern factors that have been shown to predict
levels of symptoms and care utilization as assessed at a
particular moment. In previous studies, we tested the predic-
tive power of a comprehensive symptom perception model
based on the theoretical work and empirical findings of Cioffi
[27], Kirmayer and Taillefer [28], and Pennebaker [29]. In
symptom perception theory, physical symptoms are under-
stood as the outcome of perceptual–cognitive processes
regarding input, attention, and attribution. Several symptom
perception-related variables were found to directly or indi-
rectly predict unexplained physical symptoms and/or care
utilization as measured at a particular moment, whereas the
effects of age, gender, and SES were mediated by these
variables [23,30]. In addition, the results showed that these
variables differentially affected experience (self-reported
symptoms) and presentation (GP-registered symptoms and
care utilization) of health problems [23].
The second objective of this study was to apply the
symptom perception model from our previous studies to
change in medically unexplained symptom level and fre-
quency of GP consultations thereby identifying patient-
related predictors of change. Three questions were addressed:
(1) Do variables derived from the symptom perception
model affect symptom and care change in the sequence
described in the model? (2) To what extent are the effects of
age, gender, and SES on symptom and care change medi-
ated by these variables in the predicted sequence? (3) Does
the predicted sequence of variables vary with regard to
experience and presentation?
Hypotheses
First, the following sequence of symptom perception-
related variables was hypothesized: input variables (number
of chronic physical diseases and/or employment status and/
or housekeeping) affect the experience and presentation
matic Research 57 (2004) 379–389