Date sent:      	Sun, 20 Oct 2002 

Subgroups of Fibromyalgia Patients: Evidence for Heterogeneity and an
Examination of Differential Effects Following a Community-Based Intervention

J of Musculoskeletal Pain, Vol. 10(3) 2002, pp. 9-32

Heather R. Walen, PhD; Terry A. Cronan, PhD.; Eva R. Serber, BA; Erik
Groessl, PhD; Karen Oliver, BA, PhD Candidate

Affiliations: Heather R. Walen, PhD, is Project Director, Terry A. Cronan,
PhD, is Professor, and Eva R. Serber, BA, is Graduate Student, San Diego
State University; Karen Oliver, BA, is PhD Candidate, San Diego State
University/University of California, San Diego, Joint Doctoral Program in
Clinical Psychology; Project USE, San Diego, CA.

Erik Groessl, PhD, is Project Director, University of California, San
Diego, Health Outcomes Assessment Project, 8950 Villa La Jolla Drive, Suite
2143, La Jolla, CA 92037.

Address correspondence to: Dr. Terry A. Cronan, Project USE, 6505 Alvarado
Road, San Diego, CA 92116.

The authors thank the people at Kaiser Permanente for their cooperation and
assistance in conducting this study, and all the research assistants and
participants who made this study possible. They also thank W.A. Hillix for
comments on earlier drafts.

Preparation for this article was supported by NIH grant AR-44020 and P60
AR40770.
Submitted: March 15,2001.
Revision accepted: October 29,2001.


ABSTRACT.
Objectives: The present study had three purposes:

1. to use cluster analysis to determine whether the same three clusters of
pain patients [adaptive copers, dysfunctional, and interpersonally
distressed] identified in the literature would emerge, using different
measures that assess similar constructs in a sample of fibromyalgia [FMS]
patients;

2. to validate the classification, using variables associated with FMS, but
not used in the cluster analysis; and

3. to determine whether the clusters would respond differently to a
year-long community-based intervention focused on social support and education.

Methods: Participants were 600 members of a large health maintenance
organization, primarily Caucasian [85 percent] and female [95 percent],
with an average age of 53.9 [SO = 11.4]. Participants were part of a larger
study examining the effects of social support and education on FMS symptoms
and health care use.

Results: Cluster analysis revealed three clusters that were congruent with
past research; follow-up analyses validated the uniqueness of each cluster.
There was no evidence that the intervention differentially affected the
clusters. There were a number of cluster by time interaction effects,
suggesting that both the dysfunctional and interpersonally distressed
groups improved in symptoms [e.g., pain, depression] over time, regardless
of intervention group.

Conclusions: People with FMS may fall into distinct subgroups; however, the
utility of dividing participants into these groups in planning
interventions remains unclear. Results emphasize the need to take into
account the fluctuating nature of the syndrome, as well as the statistical
phenomenon of regression toward the mean, when examining treatment outcomes. 

KEYWORDS. Fibromyalgia, pain, clusters, intervention


INTRODUCTION

Fibromyalgia syndrome [FMS] is a chronic, often debilitating, condition
that is resistant to treatment. Its diagnosis relies on a patient's report
of chronic pain, having pain in 11 of 18 tender points, and ruling out
diseases with similar symptoms (1,2). While its most prominent feature is
widespread muscular pain, other symptoms include fatigue, sleep
disturbances, morning stiffness, headaches, depression, and irritable bowel
syndrome (1).

The multifaceted expression of FMS, along with unpredictable treatment
outcomes, has led a number of researchers to propose that FMS may not be a
single disease, and/or that subgroups of FMS patients may be identifiable
(3-6). This suggests that treatment efficacy may depend on identifiable
patient characteristics.

Past research has classified chronic pain patients into meaningful groups
(5,7-9). However, these studies were subject to two main criticisms: they
used an a priori method of classification based on the assumption that
those with elevated scores on a select number of subscales were similar in
other ways, and they were primarily psychological in nature, ignoring
possible heterogeneity in variables assessing physical health domains.

In an attempt to respond to these criticisms, Turk and Rudy (10) used a
measure that contains both psychological and physical variables [the West
Haven-Yale Multidimensional Pain Inventory: MPI], in conjunction with an
empirical statistical procedure. Cluster analyses were used to minimize
within-group differences and maximize between-group differences (11).
Studies using this procedure identified three clusters: "dysfunctional,"
characterized by high levels of pain, disability, and psychological
distress; "interpersonally distressed," characterized by interpersonal
difficulties and low levels of social support; and "adaptive copers,"
characterized by low distress and disability and high levels of
self-efficacy. While a fourth cluster ["impaired"] emerged when medical
data were added to the analyses, this profile has not been replicated on
other samples or by other researchers (12). The first three clusters have
been derived and replicated on general chronic pain patients (10,13) and
those with temporomandibular disorders (14).

Using the same clustering techniques, but using separate measures of pain,
disability, and depressive symptoms, rather than the MPI, Schoefeld-Smith
et al. (4) identified two clusters among 118 FMS patients: high and low
functioning. In these studies, the clusters were found to differ on other
meaningful variables [e.g., measures of helplessness, coping, stress,
mood], adding to the validity of the classification (4,10).

Research identifying meaningful subgroups has led to speculation that such
groups may be differentially affected by treatment; however, few studies
have examined such effects (15-17). Neither McGill (15) nor Moore et al.
(16) found treatment effects dependent on clusters de- rived from the
Minnesota Multiphasic Personality Inventory [MMPI]; however, Turk et al.
(17) did detect treatment differences by cluster.

They classified a small group [N = 48] of FMS patients who participated in
an interdisciplinary treatment program. The intervention consisted of six
half-day sessions consisting of education, exercise, and
cognitive-behavioral therapy over a one-month period. They found that the
cluster types differed in outcomes. The most important difference was that
those classified as "dysfunctional" benefited more from the intervention,
reporting lower levels of depression and perceived disability at the post
assessment than the "interpersonally distressed" and "adaptive copers."
Thus, this study supported the hypothesis that some patients may derive
more benefits from an intervention than others, based on pre-existing
differences, but this study did not include a control group.

Based on intervention research with osteoarthritis patients, we designed
and implemented a year-long intervention examining the effects of social
support and education on FMS symptoms and health care use (18). Patients
were randomly assigned to a social support, a combination of social support
and education, or to a no-treatment control group. The results indicated
that there were no differential changes in the groups for health status
[including measures of FMS symptoms], or health care costs. However,
symptoms and psychological well-being improved during the study for all
groups, including the control groups (18). Because of the lack of treatment
effects, we wanted to examine whether differences would be revealed by
examining clusters of patients.

Past studies examining subgroups of pain patients have had limitations of
small sample sizes, typically averaging 100 patients or less, and few have
monitored changes over time or as the result of treatment (15-17). The
present study used a large sample of FMS patients [N = 600] and had three
purposes: 1. to use cluster analyses to determine whether the same three
clusters identified in the literature using the MMPI and MPI could be
derived from different measures assessing similar constructs; 2. to
validate the classification by examining whether there were differences
among the clusters using variables associated with FMS, but not used in the
cluster analyses; and 3. to determine whether the groups differed on
outcomes following a year-long community-based intervention focused on
social support and education. We expected that we would replicate the three
subgroups established in the literature and that these subgroups would
differ on external variables. Further, we expected that the dysfunctional
and interpersonally distressed clusters taking part in an intervention
providing social support, or social support plus education, would show
greater gains than the cluster that was coping adaptively at the baseline
assessment, or those assigned to the control group. Such a finding would
have clinical implications, suggesting that classifying patients into
subgroups and tailoring interventions might produce more favorable results
than generic treatment programs.

© 2002 by The Haworth Press, Inc. All rights reserved.


[Copies of the complete article are available for a fee from The Haworth
Document Delivery Service: 1-800-HAWORTH. E-mail address:
mailto:getinfo@haworthpressinc.com Website:
http://www.haworthpressinc.com/store/product.asp?sku=J094 ]