Progress Review

Barthel Index for Stroke Trials

Development, Properties, and Application
Terence J. Quinn, MD, MBChB (hons), BSc (hons), MRCP (UK); Peter Langhorne, PhD; David J. Stott, MD

Background and Purpose—Robust measures of functional outcome are required to determine treatment effects in stroke
trials. Of the various measures available, the Barthel index (BI) is one of the more prevalent. We aimed to describe
validity, reliability, and responsiveness (clinimetric properties) of the BI in stroke trials.
Methods—Narrative review of published articles describing clinimetric properties or use of the BI as a stroke trial end point.
Results—Definitive statements on properties of BI are limited by heterogeneity in methodology of assessment and in the
content of “BI” scales. Accepting these caveats, evidence suggests that BI is a valid measure of activities of daily living;
sensitivity to change is limited at extremes of disability (floor and ceiling effects), and reliability of standard BI
assessment is acceptable. However, these data may not be applicable to contemporary multicenter stroke trials.
Conclusions—Substantial literature describing BI clinimetrics in stroke is available; however, questions remain regarding
certain properties. The “BI” label is used for a number of instruments and we urge greater consistency in methods,
content, and scoring. A 10-item scale, scoring 0 to 100 with 5-point increments, has been used in several multicenter
stroke trials and it seems reasonable that this should become the uniform stroke trial BI. (Stroke. 2011;42:1146-1151.)
Key Words: activities of daily living 䡲 Barthel index 䡲 clinimetrics 䡲 disabilityscales 䡲 outcomes

S troke is a disabling condition, with cerebrovascular dis-

eases being the leading cause of disability in industrial
countries. Thus, efficacy of stroke interventions is often
described via measures of disability, ie, functional assess-
ment.1 Stroke assessments that focus on basic activities of
Downloaded from http://ahajournals.org by on March 24, 2019
selected journals (Stroke, Age and Ageing, Archives of Physical
Medicine and Rehabilitation) were manually searched for relevant
articles. Particular attention was given to studies describing clinim-
etrics or use of BI in stroke trials. The intention was to provide a
narrative overview and appraisal of the strengths and weaknesses of
the BI as an outcome measure for stroke trials. It should be noted that

daily living (ADL; tasks that must be performed to allow
independent living) include functional independence mea-
sure, Katz index of ADL, and the Barthel index (BI).2
BI has become a prevalent outcome measure for stroke,
with substantial supporting research.1,3 This review discusses
development and application of the BI in its many iterations,
giving particular attention to “clinimetric” properties. Clini-
metrics is the methodological discipline that focuses on
quality of clinical measurements. Scales are traditionally
assessed for validity, reliability, and responsiveness, and
these are described in turn. Such analysis is of more than
academic interest; even the best designed trials will produce
meaningless data if assessment scales used are not “clini-
metrically” fit for purpose.
Materials and Methods
The review is based on the authors’ clinical and research experience
and is informed by a search of published literature. Electronic
databases (Medline and Embase) were searched from inception to
September 2010 inclusive, using the following truncated key words:
“Barthel;” “activities of daily living;” “disability evaluation;” and
“stroke or cerebrovascular.” In addition, key reference works2,4 and
although this critique is informed by published literature, it is not a
fully comprehensive systematic review.
Results
Development of the BI
As rehabilitation became established as a medical discipline,
many scales offering objective measures of recovery were
described.5 These instruments were usually developed “in
house” and often are not subject to further assessment. In the
“chronic disease” hospitals of Baltimore, a “Maryland dis-
ability index” was developed.6 Dr Florence I. Mahoney and
Dorothea W. Barthel modified this scale to produce “a simple
index of independence, useful in scoring improvement in
rehabilitation,” ie, the BI.7
The scale described 10 tasks and was scored according to
amount of time or assistance required by the patient. Total
score was from 0 to 100, with lower scores representing
greater nursing dependency.(Figure 1). First used in approx-
imately 1955, Barthel’s eponymous scale became popular in
rehabilitation and was well-established by time of publica-
tion.6 Use of the BI spread quickly, such that it is now

Received October 26, 2010; accepted January 3, 2011.

From the Department of Academic Geriatric Medicine, Institute of Cardiovascular and Medical Services, University of Glasgow, UK.
Bo Norrving, MD, PhD, was the consulting editor for this paper.
Correspondence to Terence J. Quinn, MD, MBChB (hons), BSc (hons), MRCP (UK), Department of Academic Geriatric Medicine, Walton Building,
Glasgow Royal Infirmary, Glasgow, UK G4 0SF. E-mail Terry.quinn@glasgow.ac.uk
© 2011 American Heart Association, Inc.
Stroke is available at http://stroke.ahajournals.org DOI: 10.1161/STROKEAHA.110.598540

1146
 Quinn et al
Downloaded from http://ahajournals.org by on March 24, 2019
Figure 1. Four “Barthel indices” in common usage.
arguably the most popular ADL scale in clinical practice.3
Examples of BI assessment in studies of spinal injury, burns,
cardiac disease, rheumatoid arthritis, amputations, and frail
elderly are available.2
Although not designed for clinical trials and not specifi-
cally a stroke scale, BI has been used as a trial end point,
either singly or as part of a “global” measure, in landmark
studies of thrombolysis and acute stroke units, and now BI is
second only to the modified Rankin scale (mRS) as stroke
outcome measure of choice1 (Figure 2). BI use is international
and native language versions are described for several coun-
tries.8 However, not all non-English language versions of BI
have undergone the recommended forward-and-back transla-
tion process,9 and certain translated BI scales are thought to
be inappropriate for the target population.10
Variations and Modification to the BI
Several authors have proposed modifications to Barthel’s
original scale. These modifications have variously reordered
scale items,11 changed or expanded on definitions,12 changed
scoring,13 and added/removed items.14 Distinguishing be-
tween these BI scales is crucial, because even minor changes
to scales can produce substantial differences in scoring.15
Barthel Index for Stroke Trials 1147
It is unfortunate that many of these BI variations maintain
the descriptor “BI.” At least 4 stroke scales in common usage
are described as Barthel (Figure 1). This confuses the litera-
ture and complicates attempts at comparative or meta-analy-
sis. There is no consensus on the optimal version, although
for consistency we urge that a single version of BI is adopted
for stroke trials. The 10-item scale, scoring 0 to 100 with
5-point increments (Figure 1), has been used in several
multicenter stroke trials, and in the absence of any clearly
superior “Barthel” it seems reasonable that this should be-
come the uniform stroke trial BI. This scale is equivalent in
content to Collin and Wade’s BI (scored 0 –20);11 the change
in scoring values will not alter the other properties of the scale
and so clinimetrics also should be equivalent. Regardless of
scale chosen, it is good practice to describe the scale used and
to reference the original descriptor. In the remainder of this
article, we use the term BI to refer generically to scales based
on Mahoney and Barthel’s tool; specific alternative versions
of the scale are identified in the text.
Extended and truncated modifications of the BI are avail-
able and deserve comment because they have been used in
stroke trials. An “extended BI” comprises BI with additional
components from the functional independence measure.16
 1148 Stroke April 2011
Downloaded from http://ahajournals.org by on March 24, 2019
Figure 2. Prevalence of Barthel Index in contemporary published stroke trials.
Addition of measures pertaining to cognition, expression,
social interaction, and vision makes intuitive sense because
BI does not address these areas. However, by adding new
items, the nature of the scale is fundamentally changed and
validity cannot be assumed. Certain disciplines have taken BI
and added items specific to that field. For example, in
neurorehabilitation an early rehabilitation BI is used, com-
prising original BI and additional categories describing need
for intensive care, tracheostomy, and ventilation.17 Again,
validity of any “bespoke Barthel” is questionable unless
corresponding clinimetric assessment is offered.
Shortened versions of BI, comprising 3-, 4-, and 5-item
scales are also described.18,19 Using various techniques,
trialists have removed all but the most discriminating items in
the scale. Interestingly, the items included differ across the
truncated scales. The value of this approach in clinical
practice is open to question, because 10-item BI is already
considerably shorter than many other scales.
BI Methodology
In addition to heterogeneity in “Barthel” content, there is
further heterogeneity in the methodology used to administer
the scale. In the original description, BI was assessed through
interview and distant observation, and this remains the
standard assessment. For stroke trials, a variety of other
techniques have been used. Pertinent areas of heterogeneity
for clinical trial purposes are use of training, choice of
interview subject, and method of data collection.
An assessment using the BI should describe only what the
patient can do at time of grading. However, there may be a
temptation for assessors to adjust BI scores based on what the
patient should be able to do or would like to do. To
standardize the assessment, guidance notes for BI adminis-
tration are available,11 and in certain trials bespoke training
has been offered. However, unlike other stroke scales,20 there
is no internationally recognized training resource for BI. It
has been suggested that written guidance alone may not
improve rigor; rather, training and explicit descriptions of
scoring for each item are required.12
When BI is scored using an interview, the interviewer may
be a doctor, nurse, therapist, or professional researcher. There
are some data to suggest that profession may influence
grading, although differences between graders are modest.21
More important may be choice of interview subject, which
again may include nurse, care giver, family, or therapist. We
could find no data to recommend one interviewee over
another and, in practice, interview with several parties may be
required. However, patients probably should not be directly
interviewed. Studies have described poor validity of self-
reported BI, particularly in older22 and cognitively impaired
patients23 and both are cohorts likely to account for substan-
tial numbers in stroke trials.
 Quinn et al
Method of conducting BI interview can also vary; again,
trials have used varying methodologies assuming clinimetric
properties without robust testing. For trials, assessments
performed remote of a testing center or even without the need
for a researcher offer economic and time advantages. Litera-
ture describing telephone assessment is limited but studies
suggest telephone disability scoring may be systematically
lower than for direct interview.24 For BI based on postal
questionnaire, results have varied, although there is agree-
ment that good responses rates can be achieved.25,26 Because
studies to date have been modest in size, conducted in small
geographic areas, and have not used exclusively stroke
survivor cohorts, we still lack definitive data on utility of
these methodologies for clinical trial use.
Clinimetric Properties of the BI
The ideal outcome measure would be easy to administer,
show consistency with repeated use and multiple users, would
capture information relevant to patient and trialists, and detect
small changes over time.3 No perfect outcome measure exists
or is likely to ever exist. The relative importance of various
clinimetric properties will depend on the proposed applica-
tion. In a multicenter trial with outcome data collected at a
single time point, validity and interobserver reliability are
arguably the most important properties.
Reliability
Reliability describes measurement error associated with an
instrument; it can be assessed across several domains. Inter-
Downloaded from http://ahajournals.org by on March 24, 2019
nal consistency, traditionally measured with Cronbach ␣, is
the extent to which all items in a scale measure a single
factor. Higher values indicate greater consistency, although
very high values may signal a degree of redundancy in the
component items. For BI, internal consistency has been
described as good ( ␣ ⫽0.80 – 0.89) 13,27 to excellent
(␣⫽0.93).8
The reliability of repeated BI measures (test–retest reliabil-
ity) is important in clinical work because serial measures are
used to chart progress. For a clinical trial, such considerations
are less important because outcomes are likely to be mea-
sured during a limited number of predefined times. When
data are available, test–retest reliability of BI is usually
described as good.13 A review of several scales suggested BI
had better test–retest reliability than scales measuring ex-
tended ADL.28
Substantial literature of BI interobserver reliability (ie, do
independent observers agree on scores for a given subject) is
available and reference texts generally describe this property
as a particular strength of the scale.3 However, most studies
have used only modest numbers of raters/patients with
heterogeneity in assessment methodology and quality. Appli-
cability of these data to a contemporary multicenter trial is
questionable.
In a systematic review of BI in the elderly, consistent
findings included greater reliability at higher BI scores,
varying reliability dependent on assessor and interviewee,
and reliability varying across items of the scale.29 No equiv-
alent systematic review of BI in stroke is available. Reports of
BI reliability in stroke, described using Cohen ␬, range from
Barthel Index for Stroke Trials 1149
moderate (␬⫽0.41– 0.60) to good (␬⫽0.61– 0.80) to very good
(␬⫽0.81–1.00)8,27,28 To put these figures in context, recent
meta-analysis of mRS reliability reported scores ranging from ␬
of 0.25 to 0.95, and overall reliability was ␬ of 0.46.30
Validity
Validity describes the extent to which an instrument measures
what it purports to measure. In the absence of a “gold
standard” ADL measure, other methods to gauge validity are
required. Face validity of BI seems apparent, originally
formulated for neurological and musculoskeletal disease;7 it
is intuitive that BI would be a valid measure in stroke. The
content of BI includes those domains thought to be most
important to ADL measurement,31 although lack of measures
pertaining to communication, mood, and cognition can create
the anomalous situation in which a dependent stroke survivor
achieves a good BI score.12 Therefore, the BI is primarily of
value as a measure of physical dependency and is not of use
(other than as a basic patient descriptor) in studies that target
speech disorder (including dysphasia), depression, or cogni-
tive function.
Stroke care and rehabilitation has changed considerably in
the decades since the Barthel scale was first described.
Modifications to BI scoring guidance have recognized that
with appropriate assistive devices a degree of independence is
possible even if impairments remain.3 Thus, for example, a
stroke survivor with urinary incontinence can still score
independence if they are catheterized and can perform cath-
eter care.
Concurrent validity is suggested by close association be-
tween BI and clinical data such as amount of nursing time
required by patients,12 extent of motor loss,32 and radiological
size of infarct.33 Likewise, favorable construct validity is
suggested by close correlation with other measures of activ-
ity.5 In fact, such is the widespread popularity of BI that it is
often used as the gold standard comparator in studies of novel
ADL scales. Comparison of simultaneously collected BI and
mRS scores further proves validity, although mRS is proba-
bly superior for describing extremes of disability.34
ADL are a proxy of how the stroke survivor will function
in the home environment. Thus, ability of BI to predict return
home provides further evidence of validity. Lower BI is
associated with greater future disability, longer time to
recovery, and greater care needs to facilitate recovery.35 In
fact, BI measured at time of admission to a rehabilitation
setting may be a better predictor of return home than
“clinical” measures.33 Change in BI over a set time may be an
even more powerful predictive tool.36 The predictive utility of
early BI is not clearly demonstrated and certain authors have
argued that BI measured before day 5 after the event is
suboptimal. Although BI has reasonable prognostic utility,
across various analyses of the GAIN trials mRS was superior
for prediction of discharge destination, health care costs, and
time spent at home.37
Responsiveness
To describe improvement or deterioration, the outcome mea-
sure must be responsive to change. Across a certain range of
poststroke disability, responsiveness of BI is reasonable5 and,
 1150 Stroke April 2011
Figure 3. Hypothetical cases illustrating Barthel Index “floor” and “ceiling” effects.
with 10 graded items, BI is more sensitive to change than
other common stroke scales.38
The minimal degree of change that is thought to be
clinically significant will vary according to the trial. How-
ever, even clinically modest improvements in functioning can
have substantial meaning to patients and can be important at
a population level. Literature on the minimal clinically
important differences detected with BI suggest that a change
of ⬇2 points (BI scored 0 –20) is meaningful and beyond
measurement error.25,39
A scale should span the complete distribution of the
concept to be measured; therefore, for a stroke trial, BI should
measure and detect change across the range of possible
functional outcomes. Here, a weakness of BI becomes evi-
dent; BI is not sensitive to change at extremes of ability40,41
(Figure 3). These “floor” and “ceiling” effects limit utility of
Downloaded from http://ahajournals.org by on March 24, 2019
BI and, in particular, make the scale less discriminating in
patients with severe or minor stroke events. For longer-term
assessment, BI on its own is unlikely to be sufficiently
sensitive and should be replaced or used along with other
scales. Floor and ceiling effects are not apparent with other
prevalent functional outcome measures such as mRS.
BI scale modifications designed to improve responsiveness
are described.13 Efforts to improve sensitivity are to be lauded;
however, the success of certain modifications have not been
consistently demonstrated. Changes may simply add complexity
with no other advantages.41,42 Greater detail (and hence greater
administration time) provides qualitative information for clinical
use, but this level of detail is rarely required by trialists.
Acceptability
Although we could find no studies formally exploring BI
acceptability, few would argue against acceptability of BI to
patients and assessors. Standard BI requires no direct testing and
should take only minutes, making BI among the quickest of the
ADL instruments. Time required for testing is a major factor in
determining acceptability of a scale to therapists.43 The simplic-
ity of BI makes it particularly suited to clinical trials; however,
even this relatively quick tool may present too great a burden in
practice. In the U.K. National Stroke Audit, completion rate of
BI measures was only ⬇60%.44 For novel scales, it is now good
practice to include scale “subjects” in the development process;
stroke survivor views were not used to inform the original BI or
any of its iterations. However, this potential criticism is of less
importance to an ADL scale than scales measuring societal
participation or quality of life.
Interpreting BI Data for Stroke Trials
Statistical manipulation of BI data are problematic. The
ordinal nonhierarchal nature of the scale invalidates many
“standard” comparative tests. The use of total BI score for
analysis assumes that all items are measuring a common
domain and can be summed without weighting or standard-
ization. Whether this is true in stroke is debatable, with some
studies suggesting BI is 1-dimensional45 and others suggest-
ing that certain items do not show internal consistency.46
A common approach has been to dichotomize total BI,
defining cut-offs that represent favorable and nonfavorable
outcomes. Again, there is no standard approach, with good BI
arbitrarily defined as total scores ranging from BI of 50 to 95,
with the most prevalent cut-off point at BI ⬎95.47 It has been
suggested that key scores are BI ⬍40 (representing complete
dependence on others), BI ⬎60 (transition from complete
dependence to assisted independence), and BI ⬎85 (representing
independence with minor assistance as could be reasonably
provided in a community setting).38 Some may consider poorest
outcome after stroke to be death; unlike mRS, the BI does not
have a separate score to represent mortality.
Use of dichotomized BI has been criticized as inefficient,
making use of only part of a complete trial dataset. For
example, with a cut-off BI score of ⬎85, patients starting
with minor impairment can make clinically important recov-
ery but not have impact on trial results, whereas patients with
very low BI may recover substantially but not reach the
cut-off point. Analytical methods that measure change across
the spread of data are increasingly applied.48
Conclusions
In a previous review of ADL tools, it was commented that BI
possessed advantages of completeness, sensitivity, suitability for
statistical manipulation, and familiarity.49 Based on the literature
described, strengths of BI are widespread use and ease of
application. However, sensitivity of BI is poor across the range
of possible outcomes, particularly in minor or more severe
strokes. These floor and ceiling effects are a particular issue for
stroke trials and limit the potential utility of the scale.
Despite the limitations, BI continues to be used by trialists
and, as such, attempts to improve the clinical application of the
scale are welcome. Development of BI as a trial outcome is
hindered by the substantial heterogeneity within scales described
as Barthel and the methodologies used to administer the scale.
For consistency, we urge that a single version of BI is adopted
 Quinn et al
for stroke trials; the 10-item scale, scoring 0 to 100 with 5-point
increments, is suggested as the uniform stroke trial BI.
Although there is extensive literature related to BI, questions
regarding certain clinimetric properties remain. In particular,
data relating to the reliability of BI for stroke trials and the
optimal methodology for administration are lacking. Literature
describing clinimetrics applicable to stroke trials and methods to
improve properties of scales is emerging.20,30 There is an urgent
need for clinimetric studies using BI after stroke.
Acknowledgments
The authors are grateful to the staff of Glasgow Royal Infirmary
library for document retrieval services.
Disclosure
Dr. Quinn has previously received grant funding to explore proper-
ties of the modified Rankin scale.
References
1. Quinn TJ, Dawson J, Walters MR, Lees KR. Functional outcome
measures in contemporary stroke trials. Int J Stroke. 2009;3:200 –205.
2. Wade DT. Measurement in neurological rehabilitation. Oxford: Oxford
Medical Publications; 1992.
3. Wade DT, Collin C. The Barthel ADL Index:a standard measure of
physical disability? Int Dis Studies. 1988;10:64 – 67.
4. McDowell I. Measuring health. A guide to rating scales and question-
naires. Oxford: Oxford University Press; 2006.
5. Granger CV, Dewis LS, Peters NC, Sherwood CC, Barrett JE. Stroke
rehabilitation: analysis of repeated Barthel index measures. Arch Phys
Med Rehabil. 1979;60:14 –17.
6. Wylie CM, White BK. A measure of disability. Arch Environ Health.
1964;8:834 – 839.
7. Mahoney FI, Barthel DW. Functional evaluation:the Barthel Index.
Downloaded from http://ahajournals.org by on March 24, 2019
Maryland State Med J. 1965;14:61– 65.
8. Leung SO, Chan CC, Shah S. Development of a Chinese version of the
modified Barthel Index. Clin Rehabil. 2007;21:912–922.
9. Guillemin F, Bombardier C, Beaton D. Cross-cultural adaptation of health
related quality of life measures: literature review and proposed guidelines.
J Clin Epidemiol. 1993;46:1417–1432.
10. Ali SM, Mulley GP. Is the Barthel scale appropriate in non-industrialized
countries? Disabil Rehabil. 1998;20:195–199.
11. Collin C, Wade DT, Davis S, Horne V. The Barthel ADL index: a
reliability study. Int Disabil Studies. 1988;10:61– 63.
12. Novak S, Johnson J, Greenwood R. Barthel revisited:making guidelines
work. Clin Rehabil. 1996;10:128 –134.
13. Shah S, Vanclay F, Cooper B. Improving the sensitivity of the Barthel
Index for stroke rehabilitation. J Clin Epidemiol. 1989;42:703–709.
14. Fortinsky RH, Granger CV, Seltzer GB. The use of functional assessments in
understanding home care needs. Med Care. 1981;19:489–497.
15. PIcavet HS, van den Bos GA. Comparing survey data on functional
disability:the impact of some methodological differences. J Epidemiol
Community Health. 1996;50:86 –93.
16. Prosiegel M, Bottger S, Schenk T. Der erwertiertr Barthel Index
(EBI)-eine neue skala zur erfassung von fahigkeitsstorungen bei neurolo-
gischen patieneten. Neurol Rehabil. 1996;1:7–13.
17. Schonle PW. The early rehabilitation Barthel Index. Rehabil. 1995;34:
69 –73.
18. Ellul J, Watkins C, Barer D. Estimating total Barthel scores from just
three items: the European Stroke Database “minimum dataset” for
assessing functional status at discharge from hospital. Age Ageing. 1998;
27:115–122.
19. Hobart JC, Thompson AJ. The five item Barthel Index. J Neurol Neu-
rosurg Psych. 2001;71:225–230.
20. Quinn TJ, Lees KR, Hardemark HG, Dawson J, Walters MR. Initial
experience of a digital training resource for modified Rankin scale
assessment in clinical trials. Stroke. 2007;38:2257–2261.
21. Richards SH, Peters TJ, Coast J, Gunnell DJ, Darlow MA, Poundsford
J. Inter-rater reliability of the Barthel ADL index:how does a researcher
compare to a nurse? Clin Rehabil. 2000;14:72–78.
22. Sinoff G, Ore L. The Barthel ADL Index: self reporting versus actual
performance in the old-old. JAGS. 1997;45:832– 836.
Barthel Index for Stroke Trials 1151
23. Ranhoff AH, Laake K. The Barthel ADL index: Scoring by the physician
from patient interview is not reliable. Age Ageing. 1993;23:171–174.
24. Korner-Bitensky N, Wood-Dauphinee S, Siemiatycki J. Barthel Index
information elicited overthe telephone: is it reliable? Am J Phys Med
Rehabil. 1995;74:9 –18.
25. Gompertz P, Pound P, Ebrahim S. The reliability of stroke outcome
measures. Clin Rehabil. 1993;7:290 –296.
26. Ebrahim S, Nouri F, Barer D. Measuring disability after a stroke.
J Epidemiol Community Helath. 1985;39:86 – 89.
27. Hsuh IP, Lee MM, Hsieh CL. Psychometric characteristics of the Barthel
activities of daily living index in stroke patients. J Formos Med Assoc.
2001;100:526 –532.
28. Green J, Young JA. test-retest reliability study of the Barthel Index,
Rivermead Mobility Index, Nottingham extended activities of daily living
scale and the Frenchay Activities Index in stroke patients. Disabil
Rehabil. 2001;23:670 – 676.
29. Sainsbury A, Gudrun S, Bansal A, Young JB. Reliability of the Barthel
when used with older people. Age Ageing. 2005;34:228 –232.
30. Quinn TJ, Dawson J, Walters MR, Lees KR. Reliability of the modified
Rankin Scale:systematic review. Stroke. 2009;40:3393–3395.
31. Donaldson SW, Wagner CC, Gresham GE. A unified ADL evaluation
form. Arch Phys Med Rehabil. 1973;54:175–179.
32. Wade DT, Langton-Hewer R. Functional abilities after stroke: mea-
surement, natural history and prognosis. J Neurol Neurosurg Psych.
1987;50:177–182.
33. Hertanu JS, Demoposalos JT, Yang WC, Calhoun WF, Fenigstein HA.
Stroke rehabilitation correlation and prognostic value of computerized
tomography and sequential functional assessments. Arch Phys Med
Rehabil. 1984;65:505–508.
34. Kwon S, Hartzema AG, Duncan PW, Lai SM. Disability measures in
stroke-relationship among the Barthel Index, the Functional Inde-
pendence Measure and the modified Rankin. Stroke. 2004;35:918 –923.
35. Huybrechts KF, Caro JJ. The Barthel Index and modified Rankin Scale as
prognostic tools for long term outcomes after stroke: a qualitative review
of the literature. Cur Med Res Opinion. 2007;23:1627–1636.
36. Pan SL, Wu SC, Lee TK, Chen TH. Reduction of disability after stroke
is a more informative predictor of long-term survival than intial disability
status. Disab Rehabil. 2007;29:417– 423.
37. Quinn TJ, Dawson J, Lees JS, Chang TP, Walters MR, Lees KR. Time
spent at home poststroke “home-time” a meaningful and robust outcme
measure for stroke trials. Stroke. 2008;39:231–233.
38. Dromerick AW, Edwards DF, Diringer MN. Sensitivity to changes in
disability after stroke:comparison of four scales useful in clinical trials.
J Rehabil Res Develop. 2003;40:1– 8.
39. Hsieh YW, Wang CH, Wu Sc, Chen PC, Sheu CF, Hsieh CL. Estab-
lishing the minimal clinically important difference of the Barthel Index in
stroke patients. Neurorehabil Neural Repair. 2007;21:233–238.
40. Schepers VP, Ketelaar M, Visser-Meily JM, Dekker J, Lindeman E.
Responsiveness of functional health status measures frequently used in
stroke research. Disabil Rehabil. 2006;28:1035–1040.
41. Hocking C, Williams M, Broad J, Baskett J. Sensitivity of Shah, Vanclay
and Coopers modified Barthel Index. Clin Rehabil. 1999;13:141–147.
42. Cano SJ, O’Connor RJ, Thompson AJ, Hobart JC. Exploring disability
rating scale responsiveness II: do more response options help? Neurol.
2006;67:2056 –2059.
43. McAvoy E. The use of ADL indices by occupational therapists. Br J
Occup Ther. 1991;54:383–385.
44. Irwin P, Rutledge Z, Lowe DA. Report on the national sentinel audit of
stroke. London, UK: Royal College of Physicians; 1999.
45. Laake K, Laake P, Ranhoof AH, Sveen U, Wyller TB, Bautz-Holter E.
The Barthel ADL Index: factor structure depends upon the category of the
patient. Age Ageing. 1995;24:393–397.
46. Van Hartingsveld F, Lucas C, Kwakkel G, Lindeboom R. Improved
interpretation of stroke trial results using empirical Barthel Index weights.
Stroke. 2006;37:162–166.
47. Balu S. Differences in psychometric properties, cut-off scores and
outcomes between the Barthel Index and Modified Rankin Scale in
pharmacotherapy-based stroke trials: systematic literature review. Curr
Med Res Opinion. 2009;25:1329 –1341.
48. Optimising Analysis of Stroke Trials Collaboration. Calculation of
sample size for stroke trials assessing functional outcome: comparison of
binary and ordinal approaches. Int J Stroke. 2008;3:78 – 84.
49. Gresham GE, Phillips TF, Labi ML. ADL status in stroke:relative merits
of three standard indices. Arch Phys Med Rehabil. 1980;61:355–358.