Evidence-Based Guidelines for
Using the Short Form 36 in
Cervical Dystonia
Stefan J. Cano, PhD,
1,2
Alan J. Thompson, MD,
1
Khailash Bhatia, MD,
3
Ray Fitzpatrick, PhD,
4
Thomas T. Warner, PhD,
5
and Jeremy C. Hobart, PhD
1,2
*
1
Neurological Outcomes Measures Unit, Institute of
Neurology, University College London, United Kingdom;
2
Department of Clinical Neurosciences, Peninsula Medical
School, Plymouth, United Kingdom;
3
Sobell Department of
Motor Neuroscience and Movement Disorders, University
College London, United Kingdom;
4
Department of Public
Health, University of Oxford, Oxford, United Kingdom;
5
Department of Clinical Neurosciences, Royal Free &
University College Medical School, London, United Kingdom
Abstract: We aimed to provide evidence-based guidelines
for using the Short Form 36 (SF-36) as an outcome measure
in cervical dystonia (CD). To do this, we tested the hypoth-
esized relationships between items, scales, and summary
measures of the SF-36 using psychometric analyses in data
from a postal survey of 235 people with CD. Although the
majority of subscales performed adequately, the Role Phys-
ical and Role Emotional subscales had substantial floor
and/or ceiling effects. Evidence did not support computing
SF-36 Physical and Mental Component Summary scores.
We propose guidelines that include the recommendation
that these subscale and summary scores should be reported
with caution. © 2006 Movement Disorder Society
Key words: cervical dystonia; rating scale; reliability; valid-
ity; health measurement; quality of life; psychometrics; SF-36
Patient rating scales are increasingly being used to
evaluate the impact of cervical dystonia (CD).
1
The
general recommendation is that both generic (i.e., appli-
cable across many diseases) and disease-specific (e.g.,
specifically developed for CD) measures should be
used.
2
Psychometric evidence must also be available
(i.e., reliability, validity, and responsiveness).
3
Although
this is emerging for disease-specific measures,
4
generic
scales continue to be used based on the often untested
assumption that they satisfy psychometric requirements
across different disease groups.
The Medical Outcomes Study 36-item Short Form
Health Survey (SF-36) is the most widely used generic
measure of health status
5
and has been used in dystonia.
6–8
However, recent studies have raised concerns about its use
in neurological disorders.
9 –11
As such, this study aims to
provide evidence-based guidelines for clinicians wishing to
use the SF-36 as an outcome measure in CD to ensure that
any inferences made are as valid as possible.
PATIENTS AND METHODS
Recruitment, Sample, and Data Collection
A sample of 235 people with CD (selected using a
random number generator) was recruited from a com-
plete list of 1,110 members from the Dystonia Society of
Great Britain. A booklet of questionnaires, including the
SF-36, was administered as a postal survey following
standard techniques to ensure a high response rate. The
sampling and data collection strategies are described
elsewhere.
4
Measurement Model
The SF-36 measurement model (i.e., how the authors
propose items in the SF-36 should be put into subscales, and
how subscales are scored) groups the 35 of the 36 items into
eight multi-item subscales: Physical Functioning (10 items),
Role Limitations Physical (4 items), Bodily Pain (2 items),
General Health Perceptions (5 items), Energy/Vitality (4
items), Social Functioning (2 items), Role Limitations
Emotional (3 items), and Mental Health (5 items). Subscale
scores are produced by simply summing the numbers cor-
responding to each item response. This is known as Likert’s
method of summated ratings.
12
For easier interpretation,
scores are transformed to have a range of 0 (worst health) to
100 (best health).
5
In addition, the authors recommend the computation
of two summary scores: a physical health score (Physical
Component Score; PCS) and a mental health score (Men-
tal Component Score; MCS). These are produced based
on an algorithm using U.S. general population data and
have a mean of 50 and a standard deviation of 10. The
advantage of summary scores over the eight SF-36 sub-
scale scores in clinical trials has been discussed previ-
ously
10,13
and includes more easily interpretable scores
that have better measurement precision and superior re-
sponsiveness. The process recommended for generating
SF-36 subscale and summary scores are assumptions that
must be tested.
Analyses
The SF-36 measurement model makes three funda-
mental assumptions that should be tested to determine
*Correspondence to: Dr. Jeremy C. Hobart, Department of Clinical
Neurosciences, Peninsula Medical School, Tamar Science Park, Ply-
mouth, PL6 8DH, UK. E-mail: jeremy.hobart@pms.ac.uk
Received 9 February 2006; Revised 6 July 2006; Accepted 15 July
2006
Published online 17 November 2006 in Wiley InterScience (www.
interscience.wiley.com). DOI: 10.1002/mds.21187
122 S.J. CANO ET AL.
Movement Disorders, Vol. 22, No. 1, 2007