SPINE Volume 31, Number 20, pp 2405–2407
©2006, Lippincott Williams & Wilkins, Inc.
Letters
To the Editor:
Re: Goldby LJ, Moore AP, Doust J, et al. A randomized
control trial investigating the efficiency of musculoskeletal physiotherapy on chronic back pain disorder. Spine
2006;31:1083–93.
Dr Goldby and her coauthors deserve appreciation for
their efforts in comparing different physiotherapy approaches in a 3-arm randomized controlled trial. However, their conclusion that “the spinal stabilization
program is more effective than manually applied therapy or an education booklet in treating chronic low
back pain disorder over time” (p. 1083) needs reconsideration.
The key problems in the trial are treatment dropouts
and loss to follow-up. Only 66% of randomized subjects
completed treatment. Additionally, only 62% of randomized subjects were followed up at 3 months, while
54% were followed up at both 6 and 12 months and
28% at 24 months. Inspection of Figures 3 and 4 confirms that the postrandomization exclusions (34% of those
randomized) corrupted the randomization process. This
is demonstrated by observing the markedly different
baseline scores obtained by the three groups on the pain
and disability outcomes.
Also, the losses to follow-up were quite different in the
three groups. For example, at 1 year, approximately
39% of those randomized to the education group were
followed up versus 57% of the spinal stabilization
group. These problems mean that it is not possible to
confidently attribute the observed between-group differences to treatment. Accordingly, the relative effectiveness
of these three treatments remains uncertain.
Leonardo Oliveira Pena Costa, MSc
Chris G. Maher, PhD
Jane Latimer, PhD
Back Pain Research Group
University of Sydney
Sydney, Australia
To the Editor:
Re: Goldby LJ, Moore AP, Doust J, Trew ME. A randomised controlled trial investigating the efficiency of
musculoskeletal physiotherapy on chronic low back disorder. Spine 2006;31:1083–93.
We congratulate Goldby and colleagues on their trial
of spinal stabilization exercise, manual therapy, and ed-
ucation for chronic low back pain. Their results suggest
that patients had better outcomes with the exercise
group. This paper deserves both careful consideration
and critical comment on issues that the review process
appears not to have addressed.
Better adherence to the Consolidating Standards of
Reporting Trials (CONSORT) statement1 would have
been beneficial. Issues of concern include: 1) the education booklet group had less severe clinical characteristics
(pp. 1087– 8), yet these imbalances are not adjusted for
in the analyses2; 2) analysis focuses on within-group differences and the between-group results are hidden
among more impressive within-group changes; 3) a
newly developed dysfunction score is used without detail
of its composition or sensitivity to change; 4) there are
multiple analyses, without mention of, or adjustment
for, the dangers associated with these3,4; and 5) the sample size is based on seeing a 30% difference between
groups (p. 1085) on the Oswestry Disability Index, yet
the difference between the stability and manual therapy
group was about 14% in favor of the stability group, a
difference slightly above the suggested minimally important clinical change of 10%5 but substantially lower than
the difference that the trial was powered to detect. Data
analysis was based on a per-protocol approach rather
than the widely accepted intention-to-treat approach,
which enhances internal validity; thus, the analysis had
reduced sample numbers: about 30% loss to follow-up
at 3 months and 40% at 6 and 12 months (p. 1085). Six
different subgroups of patients are analyzed, yet subgroup analyses should be predefined and carefully justified, and trials should be powered to detect subgroups
reliably.6
So what does this trial really show? Had the focus
been on the primary outcome and on between-group effects, the authors may have been less enthusiastic about
the effectiveness of spinal stability exercise. It is crucial
that the results of this trial are interpreted correctly. This
trial does not provide satisfactory evidence that spinal
stabilization exercise is more effective than the interventions with which it was compared.
Nadine E. Foster, DPhil, BSc(Hons), MCSP
Kika Konstantinou, PhD, MMACP, MCSP, MSc
Martyn Lewis, PhD, BSc(Hons)
Primary Care Musculoskeletal Research Centre
Keele University
Keele, UK
2405
2406 Spine • Volume 31 • Number 20 • 2006
Melinda Cairns, PhD, MMACP, MCSP, MSc
School of Paramedic Sciences
Physiotherapy and Radiography
Faculty of Health and Human Sciences
University of Hertfordshire
Hertfordshire, UK
References
1. Moher D, Schulz KF, Altman D. The CONSORT Statement: revised recommendations for improving the quality of reports of parallel-group randomised
trials. JAMA 2001;285:1987–91.
2. Vickers AJ, Altman DG. Analysing controlled trials with baseline and follow
up measurements. BMJ 2001;323:1123– 4.
3. Prentice R. On the role and analysis of secondary outcomes in clinical trials.
Control Clin Trials 1997;18:561–7.
4. Davis CE. Secondary endpoints can be validly analyzed, even if the primary
endpoint does not provide clear statistical significance. Control Clin Trials
1997;18:557– 60; discussion 561–7.
5. Ostelo RWJG, de Vet HCW. Clinically important outcomes in low back pain.
Best Pract Research Clin Rheum 2005;19:593– 607.
6. Rothwell PM. Subgroup analysis in randomised controlled trials: importance,
indications and interpretation. Lancet 2005;365:176 – 86.
was made. Moreover it is not clear why only women with
a history of back pain were asked to fill out more detailed
questionnaires about pain and disability and the others
were not. The authors were very positive about the reliability of the used pain provocation tests but presented a
rather selective sample of references. More critical references about those tests2 were not included.
It seems incautious to make statements about a different clinical course of PPGP and lumbar pain and present
a direction about the evaluation on generic and specific
levels in combination with pathoanatomic subgrouping
based on this study.
Caroline H. G. Bastiaenen, PhD
Janneke M. Bastiaanssen, MSc
Rob A. de Bie, PhD
Department of Epidemiology
Maastricht University
Maastricht, The Netherlands
References
To the Editor:
Re: Gutke A, Östgaard HC, Öberg B. Pelvic girdle pain
and lumbar pain in pregnancy: a cohort study of the
consequences in terms of health and functioning. Spine
2006;31:E149 –55.
In the March 2006 issue, Gutke et al reported the
findings of their cohort study on pelvic girdle pain and
lumbar pain in pregnancy. We support the authors’ assertion that the prevalence of pregnancy-related pelvic
girdle pain and the consequences in terms of health and
functioning are important issues to investigate in a cohort study. However, while reading the article, some essential doubts arose concerning the design and the underlying assumptions of this study. In the first paragraph,
the authors wrote: “PPGP occurs during pregnancy or
within the first 3 weeks after delivery.” Cohort studies
investigating study questions like those mentioned
should preferably include pregnant women during early
pregnancy and follow them for a longer period than by
the period defined by the authors themselves within
which new cases could arise. The authors indeed included women during early pregnancy, but the study
turned out to be only cross-sectional (only one measurement at early pregnancy) and not a cohort study. Subsequently, figures about prevalence and consequences of
the disorder are incomplete and only have limited value.
Secondly, the authors wrote, “PPGP and lumbar pain
have different etiologies and require different treatment
strategies” and based their study design on these assumptions. Although the reported references (not very recent)
provide the same statements, the etiology about pregnancy-related pain in the lumbopelvic region is not clear at
all.1 It seems quite probable that the assumption that
both groups are different substantially biased the results
since in the design of the study a distinction between
lumbar pain, pelvic girdle pain, and a combination group
1. Bastiaanssen JM, de Bie RA, Bastiaenen CHG, et al. A historical perspective
on pregnancy-related low back and/or pelvic girdle pain. Eur J Obstet Gynecol Reprod Biol 2005;120:3–14.
2. van der Wurff P, Hagmeijer RH, Meyne W. Clinical tests of the sacroiliac
joint: a systemic methodological review. Part 1: Reliability. Man Ther 2000;
5:30 – 6.
In Response:
Re: Gutke A, Östgaard HC, Öberg B. Pelvic girdle pain
and lumbar pain in pregnancy: a cohort study of the
consequences in terms of health and functioning. Spine
2006;31:E149 –55.
We thank Bastiaenen et al for their comments that
may help us to clarify some points in our publication.
We agree on comments concerning the design if we
strictly consider the actual publication in epidemiologically terms. The study design includes baseline in pregnancy, follow-up at 3 months postpartum, and inclusion
of the women classified with pregnancy-related pelvic
girdle pain (PPGP) into a randomized controlled trial
with a 2-year follow-up. The whole study design was not
described in the publication, which might have been of
interest.
The definition used, “PPGP occurs during pregnancy
or within the first 3 weeks after delivery,” is not as stated
by Bastiaenen et al, “the period defined by the authors
themselves within which new cases could arise” but applied from research in the field.1
We agree that in the sentence about different etiologies of PPGP and lumbar pain, “etiology” is too strong
and “clinical presentation” should have been used until
the problem is better understood. We used a large time
span (1988 –2004) of references supporting the need for
subgrouping into PPGP and lumbar pain. We wanted to
show that it has been repeated recommendations throughout the years that PPGP should be differentiated from lumbar pain. This is also the consensus of an international
Letters to the Editor • Gutke et al 2407
group of experts in the field within the European commission research (www.backpaineurope.org). The patients
with lumbar pain are mostly excluded in studies of lumbopelvic pain in pregnancy. We found it important to keep
the whole pregnant cohort within the study in order to
evaluate the lumbar pain group as well as the PPGP group
for possible similarities and differences in outcomes. The
assumption that PPGP and lumbar pain are different is a
hypothesis of the study. The aim of the study was to describe the consequences in the different subgroups not to
prove the etiology of PPGP; therefore, the comment on bias
seem inappropriate. The recommendations for research in
primary care to identify subgroups of low back pain in
order to develop specific treatment strategies are in line
with our study design.2
Bastiaenen et al are critical of our choice of references
supporting the pain provocation tests used in our study.
They refer to a review by van der Wurff et al who developed a scoring system where the authors themselves had
the highest score of all studies. This highest scored study
was in Dutch. van der Wurff et al recommended further
research in this area with an emphasis on multiple test
scores and pain provocation test of the sacroiliac joints.
We have used a multiple test score. Further conclusion
was that the Gaenslen’s and the thigh thrust tests seem to
be reliable. Both tests were used in our study and either
or both of the tests were positive in all women with PPGP
except in one with combined PPGP and lumbar pain.
There is still a need for further studies to understand and
develop valid and reliable classification of these groups.
Bastiaenen et al questioned that only women with a history of back pain were asked to fill out a more detailed
View publication stats
questionnaire. The questions regarding previous lumbopelvic pain and its consequences (sick leave due to back pain,
whether the previous back pain was pregnancy-related,
etc.) can only be answered by those who have had any
experience of lumbopelvic pain. In the final statement in
our publication, “This study shows that it is useful to evaluate on both specific and generic levels in combination with
pathoanatomic subgrouping for the understanding of the
different clinical course of PPGP and LP” the use of clinical
“course” aimed at the clinical experience from this patient
groups as well as the published result that most women
with PPGP recover postpartum3 while women with non–
pregnancy-related low back pain experience a recurrent
course.4 We find it interesting that, when using existing
clinical testing, there was a difference in self-reported function and health among those subgroups.
Annelie Gutke, RPT
Hans Christian Östgaard, MD, PhD
Birgitta Öberg, PhD
Department of Health and Society
Linköping Universitet
Linköping, Sweden
References
1. Mens JM, et al. Understanding peripartum pelvic pain: implications of a
patient survey. Spine 1996;21:1363–9; discussion 1369 –70.
2. Borkan JM, et al. A report from the Second International Forum for Primary
Care Research on Low Back Pain: reexamining priorities. Spine 1998;23:
1992– 6.
3. Albert H, Godskesen M, Westergaard J. Prognosis in four syndromes of pregnancy-related pelvic pain. Acta Obstet Gynecol Scand 2001;80:505–10.
4. Von Korff M, Saunders K. The course of back pain in primary care. Spine
1996;21:2833–7; discussion 2838 –9.