Measuring change in disability after inpatient

rehabilitation: comparison of the responsiveness of

the Barthel Index and the Functional
Independence Measure
J J M F van der Putten, J C Hobart, J A Freeman, A J Thompson
Abstract
BackgroundThe importance of evaluat-
ing disability outcome measures is well
recognised. The Functional Independence
Measure (FIM) was developed to be a
more comprehensive and sensitive
measure of disability than the Barthel
Index (BI). Although the FIM is widely
used and has been shown to be reliable
and valid, there is limited information
about its responsiveness, particularly in
comparison with the BI. This study com-
pares the appropriateness and responsive-
ness of these two disability measures in
patients with multiple sclerosis and
stroke.
MethodsPatients with multiple sclerosis
(n=201) and poststroke (n=82) patients
undergoing inpatient neurorehabilitation
were studied. Admission and discharge
scores were generated for the BI and the
three scales of the FIM (total, motor, and
cognitive). Appropriateness of the meas-
ures to the study samples was determined
by examining score distributions, oor
and ceiling eVects. Responsiveness was
determined using an eVect size calcula-
tion.
ResultsThe BI, FIM total, and FIM
motor scales show good variability and
have small oor and ceiling eVects in the
study samples. The FIM cognitive scale
showed a notable ceiling eVect in patients
with multiple sclerosis. Comparable eVect
sizes were found for the BI, and two FIM
scales (total and motor) in both patients
with multiple sclerosis and stroke pa-
tients.
ConclusionAll measures were appropri-
ate to the study sample. The FIM cognitive
scale, however, has limited usefulness as
an outcome measure in progressive multi-
ple sclerosis. The BI, FIM total, and FIM
motor scales show similar responsiveness,
suggesting that both the FIM total and
FIM motor scales have no advantage over
the BI in evaluating change.
(J Neurol Neurosurg Psychiatry 1999;66:480484)
Keywords: Barthel Index; Functional Independence
Measure; responsiveness; rehabilitation
Measuring the eVectiveness of clinical inter-
ventions by using standardised measurement
instruments is now widely accepted as being
central to good clinical practice.
1
As the
number of potential healthcare interventions
has increased disproportionately to the health-
care budget, pressure has been put on services
to show that they provide high quality care that
is cost eVective.
Measuring the outcome of healthcare inter-
ventions is a central component of determining
therapeutic eVectiveness and, therefore, the
provision of evidence-based health care. How-
ever, information generated by outcome stud-
ies is only meaningful if the measures used are
clinically useful and scientically sound.
2
Con-
sequently, it must be shown that instruments
measure the outcome under study in a way that
is reliable and valid. In addition, instruments
used for evaluative studies must also be shown
to be able to detect clinically signicant change
in the outcome measured. This property is
known as responsiveness.
3
As rehabilitation is a labour intensive and
costly intervention, evaluating its therapeutic
eVectiveness is particularly important. Al-
though some studies have shown that rehabili-
tation is benecial,
49
there is no consensus as
to which outcomes should be measured. Reha-
bilitation aims to improve various aspects of a
patients lifefor example, disability, handi-
cap, and quality of lifeand ideally these
should all be included in the outcome
assessment.
10
Despite a move towards quality
of life and patient based outcome measures,
observer rated generic measures of disability
are still widely used.
11
The skills involved in self
care and mobility are assumed to be basic to
higher levels of functioning,
12
thus improve-
ments in disability are likely to have consider-
able impact on a persons level of handicap and
health related quality of life.
The Barthel Iindex (BI) and the FunctionaI
Independence Measure (FIM) are probably
the most widely used generic disability meas-
ures. The BI was developed in 1955 as a simple
index of independence useful in scoring
disability.
13
However, it was regarded as being
too crude, too simple, and not responsive
enough to evaluate disability outcomes in
rehabilitation. Consequently, the FIM was
developed between 1984 and 1987. The
specic aims of the developers of the FIM were
to produce an instrument that provided
comprehensive and sensitive disability
measurement.
14
The FIM contains more items
than the BI, includes cognitive items, and has
more response categories.
J Neurol Neurosurg Psychiatry 1999;66:480484 480
Institute of Neurology,
Queen Square, London
WC1N 3BG, UK
J J M F van der Putten
J C Hobart
J A Freeman
A J Thompson
Correspondence to:
Professor AJ Thompson,
Institute of Neurology,
Queen Square, London
WC1N 3BG, UK. Telephone
0044 171 837 3611 ext
4152; fax 0044 171 813
6505; email
[email protected]
Received 3 March 1998 and
in revised form
5 November 1998
Accepted 11 November 1998
Although both instruments have evidence of
reliability and validity,
1520
there is only limited
information about their responsiveness. Wade
et al have suggested that the responsiveness of
the BI is adequate for clinical purposes but may
be limited in the context of research.
15
Conceptually, it is argued that the FIM is more
responsive than the BI.
12 17 21
However, empiri-
cal data supporting this claim are limited, and
those which exist use suboptimal
methodology.
18
Standard techniques for the
assessment of responsiveness, such as the
application of eVect sizes, have not been used
previously.
The aim of this study is to compare the
appropriateness and responsiveness of the BI
and the FIM in patients with multiple sclerosis
and stroke patients receiving inpatient rehabili-
tation.
Methods
STUDY SAMPLE
Patients with multiple sclerosis and stroke
patients who were admitted to the neuroreha-
bilitation unit (NRU) of the National Hospital
for Neurology and Neurosurgery between
1994 and 1997 were studied. These diagnostic
groups were studied because they have diVer-
ent clinical disease courses (acute v chronic).
The NRU is an 18 bed unit specialising in
intensive, individually tailored, goal oriented
rehabilitation of patients with neurological
disorders.
22
Patients are selected for admission
to the NRU if they have the physical potential
to actively participate in an intensive rehabilita-
tion programme; the cognitive ability to carry
over learned skills into functional tasks; and
require input from at least two disciplines other
than medical and nursing staV. Patients were
excluded from this study if their duration of
stay was less than 7 days.
MEASURES
The BI is a 10 item instrument measuring dis-
ability in terms of a persons level of functional
independence in personal activities of daily
living.
16
It is rated from observation and has
two items on a two point scale, six items on a
three point scale, and two items on a four point
scale. Item scores are summed to generate a
total score (0=minimum independence;
20=maximum independence). The BI is user
friendly and multiple studies support its
reliability and validity.
15 16 23
The FIM is an 18 item instrument measur-
ing a persons level of disability in terms of
burden of care.
14
It was developed specically
to measure functional outcomes of
rehabilitation.
20
The developers recommend
that the FIM is rated from patient observation
by the consensus opinion of a multidisciplinary
team. Each item is rated from 1 (requiring total
assistance) to 7 (completely independent).
Three independent FIM scores can be gener-
ated by summing item scores: a total score
(FIM total: 18 items), a motor score (FIM
motor: 13 items), and a cognitive score (FIM
cognitive: 5 items). Multiple studies support
the reliability and validity of FIMscales.
1821 24 25
The expanded disability status scale (EDSS)
is a multiple sclerosis specic, neurologist
rated, index grading disease severity from 0
(normal neurological examination) to 10
(death due to multiple sclerosis) in 20 steps.
26
Rating is based on the medical history and the
neurological examination. Although the EDSS
has been heavily criticised,
2729
it remains the
most widely used measure for multiple sclero-
sis due to the absence of well evaluated
superior alternatives. Evidence supports the
reliability and validity of the EDSS.
30
PROCEDURE
Patients referred for neurorehabilitation were
assessed by a senior multidisciplinary team
consisting of a neurologist, clinical nurse
specialist, occupational therapist, and physio-
therapist. Patients whom it was considered
would benet from in patient neurorehabilita-
tion had an admission date booked. On admis-
sion to the NRU patient characteristics were
recorded along with disease severity (EDSS) in
the multiple sclerosis group. For all patients,
disability measures (BI and FIM) were rated
within 96 hours of admission to, and within 48
hours of discharge fromthe NRUby consensus
opinion of a treating multidisciplinary team.
STATISTICAL ANALYSIS
Appropriateness
Appropriateness attempts to dene whether
the range of disabilities in a study sample is
similar to the range of disabilities covered by an
instrument. In this study appropriateness was
assessed by examining score ranges, means,
SDs, and oor and ceiling eVects for the BI and
three FIM scales. Mean scores indicate the
central tendency of the group, ideally these
should lie near the midpoint of the scale range.
Sample range and SD indicate the extent to
which an instrument demonstrates variability
in the study sample. The greater the variability
detected the better an instrument discrimi-
nates between subjects. Floor and ceiling
eVects, calculated as the percentage of the
sample scoring the minimum and maximum
possible scores respectively, indicate the extent
that scores cluster at the bottom and top of the
scale range. Floor and ceiling eVects represent
a limited ability to discriminate between
subjects. When an instrument measures a
restricted range of health status oor and ceil-
ing eVects indicate that the range of disability
measured by the scale is less than the range of
disability occurring in the study sample. Floor
and ceiling eVects exceeding 20% are consid-
ered to be signicant.
31
Responsiveness
Responsiveness is dened as the ability of a
measure to detect clinically important change
in the outcome of interest.
32
In this study
responsiveness was determined using an eVect
size calculation, dened as mean change score
(discharge minus admission) divided by the SD
of admission (pretreatment) scores.
33
EVect
sizes indicate, in SD units, the magnitude of
change undergone by an instrument between
two points in time. Therefore, the greater the
Measuring change in disability after inpatient rehabilitation 481
eVect size the greater the responsiveness of an
instrument. By relating change scores to the
variability of the study sample, eVect sizes
transform raw change scores with limited
meaning to a standard metric thereby allowing
comparison of diVerent instruments and diVer-
ent samples. When instruments are compared
in the same sample a direct indication of their
relative responsiveness is provided. Under
these circumstances the instrument with the
largest eVect size is considered the most
responsive.
33
In addition, paired t tests were
used to determine the statistical signicance of
disability change scores.
Results
SAMPLE
Table 1 presents the characteristics of the 283
patients studied. The multiple sclerosis group
(71% of sample) contained more women, was
slightly younger, and had a shorter length of
stay than the stroke group (29% of sample).
The EDSS scores indicated that the multiple
sclerosis group were moderate to severely disa-
bled.
APPROPRIATENESS
Table 2 presents BI and FIM score distribu-
tions for patients with multiple sclerosis and
stroke patients on admission to the NRU. For
both disease groups, patient scores on the BI,
FIM total, and FIM motor scales spanned the
entire scale range, had mean scores near the
scale midpoint, and had small oor and ceiling
eVects. These results indicate that these three
scales are appropriate to the study samples.
However, the results shown in table 2 raise
concerns over the appropriateness of the FIM
cognitive scale as a measure of cognitive
disability in the patients with multiple sclerosis
studied. Actual scores only span the upper (less
disabled) range of the scale, the mean score is
well above the scale midpoint, the SD is small,
and the ceiling eVect is only just below the rec-
ommended upper limit. The FIM cognitive
scale is, however, more appropriate to the
stroke than the multiple sclerosis sample.
RESPONSIVENESS
Table 3 presents disability change scores with
their statistical signicance and eVect sizes for
the BI and three FIM scales in patients with
multiple sclerosis and stroke patients. Change
scores for all scales in both disease groups were
positive, indicating less disability on discharge
than admission. These change scores were sta-
tistically signicant (p<0.0001) except for the
FIM cognitive score in the multiple sclerosis
group.
EVect sizes for the BI, FIM total, and FIM
motor scales were very similar in each disease
group indicating comparable responsiveness
for these three scales. Also, in both disease
groups eVect sizes for the FIM cognitive scale
were much less than for the BI, FIM total, and
FIM motor scales indicating that the FIM cog-
nitive scale is the least responsive scale.
Discussion
In this study the appropriateness and respon-
siveness of the BI and FIM were compared in
patients with multiple sclerosis and stroke
patients receiving inpatient neurorehabilita-
tion. The results show that all measures were
appropriate to the samples studied although
the FIM cognitive scale has a notable ceiling
eVect in patients with multiple sclerosis. More
importantly, the BI, FIM total, and FIM motor
scales show similar responsiveness in both dis-
ease groups.
Appropriateness of disability measures, as
dened in this study, is rarely reported in clini-
cal studies. However, when scales measure a
restricted range of health status it is important
to show the appropriateness of this range to the
study sample. The patients in this study had
moderate to severe disability as measured by
the EDSS. However, the range of cognitive
dysfunction measured by the FIM cognitive
Table 1 Characteristics of patients with multiple sclerosis and stroke patients
Disease group n
Male
sex
(%)
Age (y) DOS (days)
Admission EDSS
score range Mean (SD) Range Mean (SD) Range Median (SD)
Multiple sclerosis 201 30.8 45 (11.2) 22-73 21.4 (11.0) 10-100 7.0 (1.0) 5.0-9.0
Primary progressive 35 28.6 47 (11.9) 23-73 26.0 (15.7) 10-31 8.0 (0.9) 5.5-9.0
Secondary progressive 166 31.3 44 (11.1) 22-73 20.5 (9.5) 10-100 7.0 (0.5) 5.0-9.0
Stroke 82 53.0 52 (16.9) 19-87 49.5 (35.1) 9-148 - -
DOS=duration of stay.
Table 2 BI and FIM scores on admission: sample range, mean, oor, and ceiling eVect in
two disease groups
Scale range n Sample range
Admission score
mean (SD)
Floor eVect
n (%)
Ceiling eVect
n (%)
BI (0-20)
MS 201 0-20 12.0 (5.7) 3 (1.5) 11 (5.5)
Stroke 82 0-20 11.4 (5.5) 1 (1.2) 7 (8.5)
FIM total (18-126)
MS 201 24-122 89.4 (23.0) 0(0) 0 (0)
Stroke 82 21-123 82.5 (26.8) 0 (0) 0 (0)
FIM motor (13-91)
MS 201 13-88 59.1 (20.4) 4 (2.0) 0 (0)
Stroke 82 13-91 56.1 (20.9) 1 (1.2) 1 (1.2)
FIM cognitive (5-35)
MS 201 11-35 30.3 (5.1) 0 (0) 36 (17.9)
Stroke 82 5-35 26.4 (7.9) 1 (1.2) 11 (13.4)
MS=multiple sclerosis; oor (ceiling) eVects are the percentage of patients scoring the minimum
(maximum) possible scores.
Table 3 Comparison of BI and FIM change scores, p values, and eVect sizes
Disease groups Change score mean (SD) p Value EVect size
MS patients
BI 2.1 (2.4) < 0.0001 0.37
FIM total 6.9 (8.3) < 0.0001 0.30
FIM motor 6.9 (7.2) < 0.0001 0.34
FIM cognitive 0.1 (2.9) 0.961 (NS) 0
Stroke patients
BI 5.2 (4.4) < 0.0001 0.95
FIM total 21.9 (19.0) < 0.0001 0.82
FIM motor 19.1 (16.1) < 0.0001 0.91
FIM cognitive 2.8 (4.8) < 0.0001 0.61
EVect size=change score/standard deviation of admission score.
482 van der Putten, Hobart, Freeman, et al
scale was restricted in these patients suggesting
that this scale has limited usefulness for the
measurement of cognitive disability in patients
with multiple sclerosis undergoing neuroreha-
bilitation. Even for the stroke patients the ceil-
ing eVect of the FIM cognitive scale is notable
(13.4%) raising some concerns over its use in
this patient group.
Patients in this study are not necessarily rep-
resentative of multiple sclerosis or stroke
patients undergoing neurorehabilitation. They
were not randomly selected, and severely
cognitively impaired patients were not repre-
sented as reasonable cognition was one of the
selection criteria for admission to the unit.
These considerations underlie the need to
examine the appropriateness of scales to a
study sample.
The most important nding of this study is
the demonstration that the BI, FIM total, and
FIM motor scales have similar responsiveness.
This is perhaps surprising as the FIM was
developed specically to be more sensitive to
change (responsive) than the BI,
12 14 17
and has
more items and more response categories. The
ndings of this study suggest that the FIM has
no advantages over the BI in evaluating
changes in disability due to therapeutic inter-
ventions. This has important clinical implica-
tions as the BI is quicker and simpler to rate. In
addition, it can be rated by any healthcare
professional whereas the developers of the FIM
recommend rating by consensus opinion of a
multidisciplinary team after a period (up to 72
hours) of patient observation. Furthermore,
the BI can be administered by self report, add-
ing to its impact on the design and cost of
clinical studies.
34 35
Examining relative responsiveness is impor-
tant as it helps clinicians to choose between
competing disability measures on an empirical
basis. The more responsive a disability meas-
ure, the more useful it is for evaluative studies
as the importance of responsiveness lies in the
trade oV between sample size and statistical
power.
36
For a given sample size, using a more
responsive instrument increases the possibility
of detecting a statistically signicant result.
Similarly, for a given statistical power a smaller
sample size can be used if a more responsive
instrument is employed.
There is no consensus as to which of the
many methods of reporting responsiveness
should be used. The eVect size statistic used in
this study is widely used and recommended.
37
However, diVerent studies often use diVerent
statistical methods, thereby complicating com-
parative data interpretation. Furthermore, the
responsiveness of instruments seems to be dis-
ease dependent. In this study eVect sizes for BI
and all FIM scales are greater for stroke
patients than for patients with multiple sclero-
sis suggesting that these instruments are more
responsive in stroke patients. Consequently,
examining the responsiveness of competing
instruments in the same samples undergoing
the same interventions provides the best
indication of relative responsiveness.
In conclusion, these results show that the BI,
FIM total, and FIM motor scales have a simi-
lar ability to detect change in disability in a
selected sample of multiple sclerosis and stroke
patients undergoing neurorehabilitation. All
measures were shown to be very appropriate to
the study sample, although concerns are raised
about using the FIM cognitive scale in patients
with multiple sclerosis.
We thank medical, nursing and all therapy staV at the NRU for
their involvement.
1 Sackett DL, Rosenberg WMC, Gray JAM, et al. Evidence
based medicine: what it is and what it isnt. BMJ
1996;312:712.
2 Fleiss JL. The design and analysis of clinical experiments. New
York: Wiley, 1986.
3 Kirshner B, Guyatt G. A methodological framework for
assessing health indices. J Chron Dis 1985;38:2736.
4 Dodds TA, Martin DP, Stolov WC, et al. A validation of the
functional independence measurement and its perform-
ance among rehabilitation inpatients. Arch Phys Med Reha-
bil 1993;74:5316.
5 Carey RG, Seibert JH, Posavac EJ. Who makes the most
progress in inpatient rehabilitation? An analysis of func-
tional gain. Arch Phys Med Rehabil 1988;69:33743.
6 Kidd D, Howard RS, LosseV NA, et al. The benet of inpa-
tient neurorehabilitation in multiple sclerosis. Clin Rehabil
1995;9:198203.
7 Greenspun BG, Stineman M, Agri R. Multiple sclerosis and
rehabilitation outcome. Arch Phys Med Rehabil 1987;68:
4347.
8 Aisen ML, Sevilla D, Fox N. Inpatient rehabilitation for
multiple sclerosis. J Neurol Rehabil 1996;10:436.
9 Freeman JA, Langdon DW, Hobart JC, et al. The impact of
inpatient rehabilitation on progressive multiple sclerosis.
Ann Neurol 1997;42:23644.
10 Gompertz P, Pound P, Ebrahim S. The reliability of stroke
outcome measures. Clin Rehabil 1993;7:2906.
11 Patrick DL, Deyo RA. Generic and disease-specic
measures in assessing health status and quality of life. Med
Care 1989;27(suppl):S21732.
12 Keith RA, Granger CV, Hamilton BB, et al. The functional
independence measure: a new tool for rehabilitation. In:
Eisenberg M, Grzesiak R, eds. Advances in clinical rehabili-
tation. New York: Springer Verlag, 1987:618.
13 Mahoney FI, Barthel DW. Functional evaluation: the
Barthel index. Maryland State Medical Journal 1965;14:61
5.
14 Granger C, Hamilton B, Keith R, et al. Advances in
functional assessment for medical rehabilitation. Topics in
Geriatric Rehabilitation 1986;1:5974.
15 Wade DT, Collin C. The Barthel ADL index: a standard
measure of disability? International Disability Studies
1988;10:647.
16 Collin C, Wade DT, Davis S, et al. The Barthel ADL index:
a reliability study. International Disability Studies 1988;10:
613.
17 Hamilton B, Granger C, Sherwin F, et al. A uniform national
data system for medical rehabilitation. In: Fuhrer MJ, ed.
Rehabilitation outcomes: analysis and measurement.
Baltimore: Brookes, 1987:13747.
18 Kidd D, Stewart G, Baldry J, et al. The functional
independence measure: a comparative validity and reliabil-
ity study. Disabil Rehabil 1995;17:1014.
19 Hobart JC, Lamping DL, Freeman JA, et al. Measuring dis-
ability in multiple sclerosis: reliability of the functional
independence measure [abstract]. J Neurol 1996;243(6
suppl 2):S32.
20 Hamilton BB, Laughlin JA, Fielder RC, et al. Interrater reli-
ability of the 7-level functional independence measure
(FIM). Scand J Rehabil Med 1994;26:11519.
21 Granger CV, Cotter ACR, Hamilton BB, et al. Functional
assessment scales: a study of persons with multiple sclero-
sis. Arch Phys Med Rehabil 1990;71:8705.
22 Freeman JA, Playford ED, Nicholas RS, et al. A neurological
rehabilitation unit: audit of activity and outcome. J R Coll
Physicians Lond 1996;30:216.
23 Hobart J, Lamping D, Freeman J, et al. Measuring
neurologyis bigger better? Comparative measurement
properties of the functional independence measure (FIM)
and the Barthel index (BI) [abstract]. Neurology 1997;48(3
suppl):A235.
24 Granger CV, Cotter AC, Hamilton BB, et al. Functional
assessment scales: a study of persons after stroke. Arch Phys
Med Rehabil 1993;74:1338.
25 Hobart J, Langdon D, Lamping D, et al. Can cognitive dis-
ability in multiple sclerosis be measured from behavioural
observation? Validity of the functional independence meas-
ure cognitive scale (FIM-c) [abstract]. Multiple Sclerosis
1997;3:268.
26 Kurtzke J. Rating neurological impairment in multiple
sclerosis: an expanded disability status scale (EDSS). Neu-
rology 1983;33:144452.
27 Willoughby EW, Paty DW. Scales for rating impairment in
multiple sclerosis: a critique. Neurology 1988;38:17938.
28 Willoughby E. Impairment in multiple sclerosis . Multiple
Sclerosis Management 1995;2:1316.
29 Sharrack B, Hughes RAC. Clinical scales for multiple scle-
rosis. J Neurol Sci 1996;135:19.
Measuring change in disability after inpatient rehabilitation 483
30 Hobart JC, Lamping DL, Freeman JA, et al. Reliability,
validity, and responsiveness of the Kurtzke expanded
disability status scale (EDSS) in multiple sclerosis [ab-
stract]. J Neurol Neurosurg Psychiatry 1997;62:212.
31 Holmes W, Shea J. Performance of a new, HIV/AIDS-targeted
quality of life (HAT-QoL) instrument in asymptomatic sero-
positive individuals. Qual Life Res 1997;6:56171.
32 Guyatt G, Walter S, Norman G. Measuring change over
time: assessing the usefulness of evaluative instruments. J
Chron Dis 1987;40:1718.
33 Kazis LE, Anderson JJ, Meenan RF. EVect sizes for
interpreting changes in health status. Med Care 1989;27(3
suppl):S17889.
34 Gompertz P, Pound P, Ebrahim S. A postal version of the
Barthel index. Clin Rehabil 1994;8:2339.
35 Hobart JC, Lamping DL, Thompson AJ. Measuring
disability in neurological disease: validity of the self-report
Barthel index [abstract]. J Neurol 1996;243(suppl 2):S25.
36 Liang MH, Larson MG, Cullen KE, et al. Comparative
measurement eYciency and sensitivity of ve health status
instruments for arthritis research. Arthritis Rheum 1985;28:
5427.
37 Scientic advisory committee of the Medical Outcomes
Trust. Instrument review criteria. Medical Outcomes Trust
Bulletin 1995;3:I-IV.
484 van der Putten, Hobart, Freeman, et al