(Stroke. 1998;29:1737-1739.)
© 1998 American Heart Association, Inc.
Trials of Community Rehabilitation Need To Be of Adequate Sample Size
Charles Wolfe, MD
Division of Public Health Sciences,
United Medical and Dental Schools
Anthony Rudd, FRCP
Department of Care of the Elderly,
Guy's and St Thomas' Hospital Trust
Kate Tilling, MSc
Division of Public Health Sciences,
United Medical and Dental Schools,
London, UK
To the Editor:
Evaluations of the most cost-effective ways of providing rehabilitation
after stroke are central to planning stroke care. Although reporting of
good-strength, randomized controled trial evidence is necessary in the
era of evidence-based medicine, basic principles have to be adhered to.
Both the trials by Holmqvist and colleagues1 and
Rogers2 are useful pilot studies but are statistically
weak and do not provide conclusive evidence for clinicians and health
care planners on how to provide care.
In the Swedish trial1 the results are applicable to less
than 10% of stroke patients. The main outcome measure for the trial is
not specified, and the sample size for detecting a specific difference
for this main measure is not detailed. The authors state that 130
patients would be required to detect an unspecified difference in
costs; however, the trial includes only 81 patients. Hence, like the
evaluation by Rodgers,2 the results must be considered as
pilotd data. There is not sufficient power to detect important clinical
differences in outcome, and the authors do not discuss the fact that
the nonsignificant differences could, with an adequate sample size,
become negative outcomes in a larger study.
In a similar trial of early discharge from hospital to a community
rehabilitation team in London, UK, we randomized stroke patients in
hospital and followed them up for 1 year. There were no significant
clinical differences between the groups at 1 year, but the early
discharge from hospital to a rehabilitation team option has been shown
to be effective.3 This trial required 260 patients to have
sufficient power to detect clinically significant differences in
Barthel score at 1 year.
If trials are to undertaken, we as clinical researchers have an
obligation to ensure that hypotheses can be answered, and this requires
rigorous trial design, with adequate numbers of patients to detect
differences in outcome should they exist.
References
1.
Holmqvist LW, von Koch L, Kostulas V, Holm M,
Widsell G, Tegler H, Johansson K, Almazán J, de Pedro-Cuesta J. A
randomized controlled trial of rehabilitation at home after stroke in
southwest Stockholm. Stroke. 1998;29:591–597.[Abstract/Free Full Text]
2.
Rodgers H, Soutter J, Kaiser W, Pearson P, Dobson R,
Skilbeck C, Bond J. Early supported hospital discharge following acute
stroke: pilot study results. Clin Rehabil.. 1997;11:280–287.[Abstract/Free Full Text]
3.
Rudd AG, Wolfe CDA, Tilling K, Beech R. The
effectiveness of a package of community care on one year outcome of
stroke patients. BMJ. 1997;315:1039–1044.[Abstract/Free Full Text]
Response:
Lotta Widén Holmqvist;
Senior Lecturer;
PhD Lena von Koch;
RPT;
PhD candidate
Division of Neurology,
Karolinska Institute,
Huddinge University Hospital,
and the Department of Physical Therapy,
Karolinska Institute,
Stockholm, Sweden
Vasilios Kostulas, MD, PhD
Division of Neurology,
Karolinska Institute,
Huddinge University Hospital,
Stockholm, Sweden
Margareta Holm;
RPT, MSc
Department of Physical Therapy,
Huddinge University Hospital,
Stockholm, Sweden
Gunilla Widsell;
OT
Department of Occupational Therapy,
Huddinge University Hospital,
Stockholm, Sweden
Helena Tegler;
SLP
Department of Geriatric Medicine,
Huddinge University Hospital,
Stockholm, Sweden
Kerstin Johansson;
SLP
Division of Neurology,
Karolinska Institute,
Huddinge University Hospital,
Stockholm, Sweden
Javier Almazán, RN
Department of Applied Epidemiology,
National Centre of Epidemiology,
Carlos III Institute of Health,
Madrid, Spain
Jesus de Pedro-Cuesta, MD, PhD
Division of Neurology,
Karolinska Institute,
Huddinge University Hospital,
Stockholm, Sweden,
and the Department of Applied Epidemiology,
National Centre of Epidemiology,
Carlos III Institute of Health,
Madrid, Spain
The excellent publication of Rudd and colleagues1
published simultaneously with ours2 and that
of Rodgers et al3 add to the growing recognition that
early supported discharge with continuity of rehabilitation at home can
be feasible in combination with a considerable reduction in the use of
bed-days for stroke patients.
We read with interest the comments by Rudd and colleagues and agree on
the importance of rigorous trial design with adequate sample size in
trials of community rehabilitation. We did not state in our article, as
was mentioned in the letter of Rudd and colleagues, that 130 patients
would be required to detect differences in costs. The fact that our
trial had a twin purpose, namely, to determine whether our model of
home rehabilitation was (1) more effective and/or (2) resource
efficient, rendered calculation of study size complex. A detailed
description of power calculations in our trial has been published
elsewhere.4 Any potential differences in savings averaging
under 7000 Swedish Krona (SKr) per patient were considered irrelevant
if differences in effect or patient satisfaction with care were
minimal. On the other hand, differences in outcome of less than 40%
(closely equivalent to an odds ratio of 1.5) may be questioned because
of limitation in sensitivity, reliability of instruments, and in
general, validity. As seen from the differences in study size
calculated on the basis of different assumptions listed in the
Table
, the power requirements for
demonstrating differences in cost were considerably lower than those
for showing statistically significant differences in effect. We
concluded that a study size of 130 patients would allow for indication
of possible moderate positive effects and demonstrate important
differences in secondary effects (odds ratio, >3), as well as savings
of a magnitude that could motivate changes in healthcare policy, and
thus yield an acceptable balance of results. We acknowledge that our
trial included only 81 patients and hence has the power to detect
differences in cost for utilization of healthcare resources and not
clinical outcome. In this regard, we did not exclude the possibility of
pooling our results with those from other planned or ongoing studies in
comparable European populations, and we therefore welcome
meta-analysis.
As mentioned by Rudd and colleagues, we do not specify one main outcome
measure. To our knowledge, there is no consensus on appropriate main
outcome measure for studies that focus on moderately disabled stroke
patients. As in the study by Rodgers et al,3 we used
several main outcome measures to capture possible effects on
impairment, disability, and/or handicap level. Several
authors5 6 have recently pointed out that the assessment
of stroke disability should take into account not only the patients'
ability to perform basic or instrumental ADL (eg, the Barthel ADL
index) but also the patients' perceptions of their emotional, social,
and physical functions and the ease with which they are performed. The
battery of stroke disability measurements chosen for our trial fulfills
such requirements and has subsequently been recommended by authorities
in the field.5 6
In a rigorous trial design of community rehabilitation, it is
important, in our opinion, to reduce the discrepancies in initial
medical attention, care, and rehabilitation. A stroke unit is thus far
the only known organization producing services for which an impact on
mortality, long-term care, and the level of dependence in ADL has been
demonstrated.7 Unlike those in other
studies,1 3 all of our patients received similar initial
medical attention and early rehabilitation at the Department of
Neurology at Huddinge University Hospital, organized as a stroke
unit.2 Thus, when the admittance procedures for stroke
patients were altered at the Huddinge Hospital and not all patients
from the catchment area received care at the Department of Neurology,
we were no longer able to recruit all stroke patients from the
population (as was previously the case) nor were we able to ensure
similar initial medical attention, care, and rehabilitation. A nursing
strike, a physical therapy strike, and periods of shortage of speech
therapists were other factors beyond our control that reduced the
number of patients included in the study.
Rudd and colleagues point out that our results2 are
applicable to less than 10% of the stroke population. In our opinion,
it is not likely that there will be one best way of organizing the
rehabilitation service that will be feasible for such a
heterogenous group as stroke patients. Our
study2 focused on moderately disabled stroke patients who,
in line with prior experience, are those who might benefit most from
specific interventions.8 9 Our patient selection criteria,
based on level of cognitive function and ADL capacity at 1 week, would
strengthen the appropriateness of our stroke population10
for the purposes of experimental evaluation research into
rehabilitation techniques. Nevertheless, the results of our
study2 should not be used for planning stroke services for
the more severe or unspecified type of stroke
patients.10
References
1.
Rudd AG, Wolfe CDA, Tilling K, Beech R. Randomized
controlled trial to evaluate early discharge scheme for patients with
stroke. BMI. 1997;315:1039–1044.
2.
Widén Holmqvist L, von Koch L, Kostulas V, Holm
M, Widsell G, Tegler H, Johansson K, Almazán J, de Pedro-Cuesta
J. A randomized controlled trial of rehabilitation at home after stroke
in Southwest Stockholm. Stroke. 1998;98:591–597.
3.
Rodgers H, Soutter J, Kaiser W, Pearson P, Dobson R,
Skilbeck C, Bond J. Early supported hospital discharge following acute
stroke: pilot study result. Clin Rehabil. 1997;11:280–287.
4.
Widén Holmqvist L. Development and
Evaluation of Rehabilitation at Home After Stroke in South-west
Stockholm [thesis]. Stockholm, Sweden: Karolinska Institute;
1997.
5.
Duncan PW. Stroke disability. Phys Ther. 1994;74:399–407.[Abstract/Free Full Text]
6.
Wade DT. Measurement of Neurological
Rehabilitation. Oxford, UK: Oxford Medical Publications; 1992.
7.
Stroke Unit Trialists' Collaboration. Collaborative
systematic review of randomised trials of organised inpatient (stroke
unit) care after stroke. BMJ. 1997;314:1151–1159.[Abstract/Free Full Text]
8.
Sunderland A, Fletcher D, Bradley L, Tinson D, Langton
Hewer R, Wade DT. Enhanced physical therapy for arm function after
stroke: a one year follow up study. J Neurol Neurosurg
Psychiatry.. 1994;57:856–858.[Abstract/Free Full Text]
9.
Jeffrey DR, Good DC. Rehabilitation of the stroke
patient. Curr Opin Neurol. 1995;8:62–68.[Medline]
[Order article via Infotrieve]
10.
Kalra L, Crome P. The role of prognostic scores in
targeting stroke rehabilitation in elderly patients. J Am
Geriatr Soc. 1993;41:396–400.[Medline]
[Order article via Infotrieve]
