NII-Electronic Library Service
mp4twtz#
Speeial
ag30gca8e
431-439H
(ZO03")
Lecture
Robert
Evidence-Based
Physical
Therapy
*
D.
Herberti'2,
Catherine
Sherringtoni'3,
Anne
M
Christopher
Maheri'2
and
Mark
EIkinsL5MoseleyL4,
Abstract
Professional
practice
has
alwaysbeen
based
on evidence of one sort or another,So
whatis
new about''evidence-based"
practice
andhew
does
evidence-basedpractice
differ
from
whathealth
professionals
have
always
done?
The
novelty of evidence-basedpractice
is
that
it
emphasizes routine, systematic and criti-cal useof
high
quality
clinical researchfor
setving clinicalproblems.
Antagonists
have
raised a series of objectionsto
evidence-basedpractice
including
that
it
is
too
time-consummg,
there
is
not enough evidence, andthe
eviclenceis
notgood
enough.We
considerthese
and other objections and concludethat,
whilethere
is
sometruth
in
each, none needpreclude
evidence-basedpractice.
Difficulties
in
implementation
of evidence-basedpractice
needto
be
weighed againstthe
bene-fits
ofbasing
clinicatpractice
onthe
least
biased
forms
of evidence.Randomized
triats
and systematic reviews newprovide
us with suthcient evidenceto
makeinformed
judgments
aboutthe
effects of a range ofphysical
therapy
interventions.
The
best
evidence confirmsthe
value of some current
physical
therapy
practices
andthe
ineffectiveness
of others.We
briefiy
summarize
studies which
demonstrate
the
effectiveness ofphysical
therapy
interventiens
for
chronic musculoske}etalpain,
stress urmary incontinence, stroke, respiratorydisease
andprevention
offalls
in
the
elderly,
Key
word:physical
therapy,
evidence-based medicineThis
paper
considers sometheoretical
andpractical
issues
relatedto
implementation
of evidence-basedphys-ical
therapy.
It
begins
with abrief
overviewof
whatis
implied
by
evidence-basedpractice
anddiscusses
how
this
differs
from
traditienal
clinicalpractice.
It
then
con-siders somefrequently
raised objectionste
the
evidence-based
practice
model.The
paper
conctudesby
summa-rizing evidence of
the
effectiveness of somekey
physi-cal
therapy
interventions,
Most
ofthe
materialin
this
paper
has
been
previously
published
in
English-tanguage
*1)2)3)4)5}
pt\tsdiptv:nd<mp7rkza
Centre
forEvidence-Based
Physiotherapy,
Scheot
ofPhysio-therapy, University ef
Sydney,
AustraliaSchool
ofPhysiotherHpy,
University
ofSydney,
Australia
Prince
ofWales
Medical
ResearchInstitute,
Australia
Rehabilitation
Studies
Unlt,
Department ofMedicine.
Universityof
Sydney,
Australia
Department of
Respiratory
Medicine.
Royal
Prince
AlfredHospital,
Austraiia
Corresponclence/
Dr.
Reb
Herbert,
School
of Physiotherapy,University
ofSydney.
PO
Box
170,
Lidcombe
NSW
1825,Australia(e-mail/
R.Herbert@ths.usyd.edu.au)
journalsi'3).
What
is
Evidence-Based
Practice?
Sackett
and colleagues coneeive ofevidence-basedprac-tice
asconsisting
of afive-step
process
that
is
routinely carriedout
in
clinical encounters,The
process
involves
(i},
askinganswerable
clinicalquestions,
{ii},
finding
the
best
evidence with whichto
answerthese
questions,
(iii),
critically appraising
the
evidence,(iv),
applyingthe
evi-dence
to
clinicalproblems,
and(v),
evaluatingthe
effects ofthe
interventien
onindividuals4}.
These
five
steps alludeto
some ofthe
mostimportant
distinctions
between
evidence-basedpractice
andtradi-tional
models of cliniealpractice.
First,
evidence-basedpractice
begins
with an acknowledgment of uncertainty,This
contrasts with sometraditional
models of clinicalpractice
in
which uncertaintyis
seen as afailing.
An
attitude of itncertaintyis
likely
to
better
equiphealth
professionals
to
deal
with rapidly changing evidence.432
ge7esza4
A
secenddistinction
is
that
evidence-basedpractice
emphasizes systematic andcritical
use ef evidence4).It
involves
recordingquestions
that
arisein
clinicalprac-tice,
rankingthem
in
order ofimportance,
andthen
tack-ling
them
in
an optimal way,Evidence
is
chosen onthe
basis
of
its
probable
validity.There
is
an emphasis endeciding
if
the
intervention
willproduce
the
desired
out-comes without unreasonable risks and at a reasonable cost.This
differs
from
traditional
models ofpractice
in
whichthere
maybe
priority
given
to
elinical experience as aform
of
evidence5)6), where research evideneeis
oftenhappened
upon
ratherthan
strategically sought out, and where appraisal ofthe
quality
of clinicaL researchis
superficial ordoes
not occur at all.A
systematic approaehto
the
use
of evidencehelps
avoidthe
temptatien
to
attend onlyto
that
evidenee which supportspre-cen-ceived
ideas
of whichtherapies
are effective.An
implicit
assumption ofthe
evidence-basedpractice
modelis
that
well-cenducted clinical research oftenpro-vides
the
best
answersto
clinicalquestions.
That
is,
where
good
quality,
relevanz clinical researchis
availableit
usuallytakes
precedence
evertheory
orpersonal
expe-rience, eventhe
theories
or experiences of experts7)(but
see reference8).
The
role of clinical experience, clinical wisdom andintuition
is
primarily
in
makingbest
use ofgood
evidenceto
meetindividual
patients'
needs andpreferences.
The
requirement of "good evidence" necessarilyrestricts
the
focus
in
evidence-basedpractice
to
optima]ly
designed
clinical research,The
optimal studydesign
willdepend
onthe
clinicalquestion.
For
example,the
best
evidence aboutthe
effects
oftherapy
is
previded
by
ran-domized
trials
or systematie reviews of randomizedtri-als7)9'i3)
{but
alsosee
references14,
15
andthe
ensuingletters).
Questiens
aboutdiagnostic
tests
are usuallybest
answeredby
studiesin
whichthere
is
independent
(blind)
comparison
ofthe
tesz
with a"gold
standard"test
4)i6).Questions
aboutprognosis
arebest
answeredby
studieswhich
prospectively
monitor well-defined cohortsfrom
an early and uniformpoint
in
the
eourse oftheir
condi-tion4)i7).The
mostdifficult
questions,
those
about
patients'
beliefs
andthe
meaningsthey
attachto
their
experi-ences, maybe
best
explored with carefully conductedqualitative
researchi8)i9).Evidence-based
practice
does
notimply
that
clinicaldecisions
shouldbe
made on thebasis
of clinical researchalone.
Key
proponents
of evidence-basedhealthcare
have
emphastzed
that
the
evidenceprovided
by
cUnical research must cemplement other sorts ofinformation,
such asinformation
aboutindividual
patients'
specificneeds and
preferences4}8),
Good
clinicians are ableto
dis-es3otseese
cern
these
needs andpreferences.
In
the
best
models of evidence-basedpractiee,
evidence aboutthe
effects oftherapy
{or
accuracy ofdiagnestic
tests,
or ofprognoses)
informs,
but
does
notdominate
clinicaldecision-making.
The
physical
therapist
draws
onpast
clinical experienceto
applythe
results of researchto
the
care of individ-ualpatients.
The
best
decisions
are made with thepatient,
notin
journals
andbooks.
Objections
to
Evidence-Based
Practice
The
preceding
sectionhas
deseribed
a model of clini-ealpractice
that
probably
differs
significantlyfrom
whathappens
in
eventhe
most evidence-based clinicalset-tings.
Real-world
evidence-basedpractice
faces
significantpractical
diffcuLties,
In
addition,legitimate
philesophical
and
theoretieal
objectionshave
been
raised against mod-els of evidence-basedpractice
(see,
for
example, refer-ences20,
21).
In
this
section we attemptto
confront some ofthe
objections
to
evidence-basedpractice,
The
emphasis willbe
on objectionsto
the
useof
systematic reviews andrandomized controLled
trials
in
rnakingdecisions
abouttherapy,
Our
conclusions willbe
thatthere
are,indeed,
some seriouspracticaL
theoretical
andphilosophical
prob-lems
with evidence-basedpractice,
Nonetheless,
evidence-based
practice
offers atleast
oneprofound
advantage over a!ternative models of clinicalpractice
in
that
opti-mal useis
made efthe
least-biased
evidencefrom
clini-cal research,What
arethese
practical,
theoretical
andphilosophical
objections
to
evidence-basedpractice?
Evidence-based
practice
is
too
tiine-consuming
to
be
prac-tical
Even
withpractice
and optimal resources,the
process
of
finding
and
critically appraisingthe
best
evidenceper-taining
te
a single clinicalquestion
usuallytakes
consid-erabletime.
As
a consequence,it
is
notpractical
to
usethe
best
evidenceto
deal
with every uncertaintythat
arlsesin
every clinical encounter, and evenif
there
wasgood
quality
evidenceto
answer allclinicalquestions,
not allpractice
couldbe
evidence-based.Any
realistic modelof evidence-based
practice
mustinvotve
prioritizing
clin-ical
questiens.
Time
shou!dbe
devoted
to
answeringquestions
that
are commonly seenin
practice,
have
important
censequences,have
potential
for
eitherbene-ficia!
orharmful
treatment,
orincur
considerable cest22).How
muchtime
is
and shouldbe
spent seeking out and appraisingthe
evidence?Most
physical
therapists
spend
little
time reading clinical research23) and,because
few
physical
therapists
have
training
in
clinical appraisaLNII-Electronic Library Service
Evidence-Based
reading
time
maybe
spent sub-optimally.Rational
deter-mination of the amount of timethat
shouldbe
spentseeking out and appraising evidence requires
information
both
aboutthe
effectiveness of current clinicalpractices
and abouthow
much of animprovement
in
effectiveness couldbe
accruedin
agiven
amount oftime
by
search-ing
for
and appraisingpapers.
Unfertunately,
data
onthese
issues
are elusive.Our
viewis
that
much ofclin-ical
practice
is
far
from
optimally effective andthat
potentially
even modest amounts oftime
spentin
the
judicious
application of evidenceto
clinicaldecision
mak-ing
could substantiallyimprove
clinieal outcomes.As
just
one example, exerciseis
prescribed
with equal frequen-cyfor
acute and chroniclow
back
pain24)
but
systemat-ic
reviewsindicate
there
is
strong evidencethat
exer-cisetherapy
is
effectivefor
chrenic,but
not acute,low
back
pain25)26}.
This
suggeststhat
changesin
exerciseprescriptien
practices
could significant}yimprove
out-comesin
patients
withlow
back
pain,
We
expectthat
manypractices
would converge rapidly onthis
outcomeif
scaree
time
was usedto
answer
key
clinical
questions.
Most
clinicians arebusy.
Where
can theyfind
time
to seek and critically appraisethe
evidencefrom
ctinical research?There
are numerouspossibilities.
Time
spentin
formal
continuing educatien activities(staff
seminars,for
example) maybe
better
spentby
individuals
or smallgroups
ofphysical
therapists
answeringtheir
ownclini-cal
questions.
Depencling
on
the
clinical
setting,
case
con-ferences
eould
also
be
re-strueturedso
that
they
create
learning
experiencesfor
staff
as wellas
deal
withpatient's
problems.
These
and other suggestionshave
been
madeby
Sackett
and colleagues4).Time
spentbusi-ly
applyingineffective
orharmful
therapies
wouldbe
bet-ter
spent seeking out and critically appraisingbest
evi-Physical
Therapy
433
dence,
Secondary
sources ofinformation
{such
asthe
clinicalpractice
guidelines,
ACP
Journal
Club,
Evidence-Based
Medicine
andCritically
Appraised
Papers
in
the
Australian
Journal
ofPhysiotherapy)
distill
thekey
ings
ofhigh-quality
papers.
Consequently
they
ly
provide
a significanttime-saving
mechanismfor
busy
PractitionersZ7}2S).
7}tere
is
not enough evidenceIdeally,
atteast
from
apurely
professional
point
ofview, there would
be
good
clinical
research answeringall
important
clinicalquestions.
Of
eourse,
that
is
notthe
case.
It
has
been
claimedthat
there
is
not enoughdence
to
practice
evidence-basedphysical
therapyZ9).
How
rnuch clinical research exists andhew
rnuch canit
assistclinical
decision
making?It
ls
difficuit
to
quantify
the
volume of ciinica[ researchin
physical
therapy,
However
it
is
possible
to
estimate,at
least
rQughly,the
number of relevant randomizedals and
systematic
reviews.The
Centre
for
Based
Physicai
Therapy,
with assistancefrom,
amongothers,
the
Rehabilitation
andRelated
Therapies
Field
ofthe
Cochrane
CoLlaboration,
has
attemptedto
identify
al]randomized controlled trials and systematic reviews
in
physical
therapy
and collatethese
onthe
Physiotherapy
Evidence
Database
{PEDro/
httpi!fwww.pedro.ihs.usyd.
edu.au).
At
the
time
of
writing3,152
randomizedor
randomized
trials
and
514
systematic
reviewshad
been
identified
(see
al$o references3,
30).
More
than
200
domized
trials
and systematic reviews onPEDro
pertain
to
each ef thefolaowing
subdisciplines ofphysical
py:
cardiothoracics, continence and women"shealth,
gerontology,
musculoskeletal, neurology, orthopedics andMusculoskeletal
Gerontology
Cardiothoracics
Neurology
Orthopaedics
Sports
Women's
Health
Paediatrics
Ergonomics
Other
Fig.1
Number
available(reprinted
O
200
400
600
800
1000
Number
of
records
of randomized controtled trials and systematic reviews
in
eachphysical
therapy
sub-disciplineln
April
2002
434
mp#fftzeeg3otseese
covL-oomL.-oLooE=c
¢>-.--tusE=o
2500
2000
1500
1OOO
500
Fig.
2
o19551960
1965
19701975
1980
Year
Cumuiative
number of randomizedreviews
in
physical
therapyby
yearreference
[3},
with permission}1985
1990
1995
2000
controlled trials and systematic
{reprinted
from
Figure
1,
p,44
ofeligibility
criteria
random
allocation
concealedallocation
groups
similar atbaseline
subject
b[indjng
therapist
blinding
assessor
blinding
<
15%
dropouts
intention-to-treatanalysis
between-groupstatisticalcomparisons
point
measures
and
variability
data
Fig.
3
o
20
40
60
80
%
of
randomized
controHed
trials
Percentage
ofrated randomlzed controlledtrials
that
satisfied eachitem
of
the
PEDre
scale(reprTnted
from
Figure
4,
p.
45
ofreferenceE3],
withpermissien)
sports
(Fig.
1)3).
This
is
eneughto
tackle
many funda-mental clinicalquestions,
though
there
are notyet
enough
trials
in
most areas ofphysical
therapy
to
pro-vide convincing replication on every
permutation
efther-apy
in
every settingfor
everypatient
group.
In
some areas ofphysical
therapy,
the
volume oftrials
and reviewsis
not sufficientto
have
any realimpact
onclin-ical
practice.
However,
given
the
exponential rate efpub-lication
of clinicaltrials
and systematic reviewsin
phys-ical
therapy,
{Fig.
2)3)
this
will almost certainly changein
the
nearfuture.
It
is
like]y
that
most clinicianshave
not read allof thehigh
quality
evidencethat
pertains
to
their
own clinicalquestions.
In
this
sense atleast
thereis
an abundanceof evidence.
It
is
probably
reasonablete
expect al]prac-tising
therapists
to
be
aware ofkey
trials
and reviewsin
their
area ofpractice.
1}ie
evidenceis
notgood
enough
Certain
features
of clinicaltrials
{such
as concealmentof randomisation,
blinding
of subjects and asses$ors, and adequacy offollew-up}
tend
to
be
associated with small-er effect sizes, suggestingtrials
whichhave
these
fea-tures
tend
tobe
less
biased:Si),
Other
trials
iack
these
features,
and so we should expectthat,
on average,they
willbe
biased.
In
physical
therapy,
the
typicalrandom-ized
trial
laeks
concealment of allocation andhas
unblind-ed
patients,
assessors andtherapists,
but
does
have
ade-quate
follow-up
(Fig.
3}3),
There
mustbe
real concern aboutthe
capacity ofthe
ryplcal trialtoprovide
anunbi-ased
picture
ofthe
effects oftherapy.
Fortunately,
the
NII-Electronic Library Service
Eviclence-Based
Trials
on
PEDre
are scoredfor
methodologicalquality
and
the
medianPEDro
scorefor
randomizedtrials
in
physical
therapyhas
crept upfrom
3flO
in
the
I960s
to
its
current value of5flO.
Systematic
reviews{such
asthose
conductedby
the
Cochrane
Collaboration)
synthesizethe
findings
ofclini-cal
trials,
Ideally,
systematic reviews weuld objectively assesstrial
quality
andthen
pool
the
findings
ofhigh
quality
studiesto
provide
less
biased
and moreprecise
estimates of
the
effects oftherapy.
However,
there
are some realdifficulties
that
arise when an attemptis
madeto
systematically review clinicaltrials
in
all areas ofhealth
care.Three
suchpreblems
arediscussed
below.
The
first
two
issues
are a]so relevantte
readers of indi-vLdual clinicaltrlals.
L
Pub]ication
bias,
This
is
the
bias
that
arisesbecause
trials
withpositive
findings
are morelikely
to
be
pubEshed
than
trials
with negativefindings.
A
sequence
is
that
positive
studies are morelikely
to
be
reviewed, and reviews arelikeLy
to
contained estimates of
treatment
effects32),Although
it
is
often assumed
that
exhaustive searching reducesthe
potentiat
for
publication
bias,
it
is
possibie
that
this
actually
increases
thepotentiul
for
bias
by
ing
lower
quality
studiesin
the
review33).There
arecurrently no completely satisfactery solutions
te
the
problem
ofpttblication
biasU4}.
2.
Scoring
of studyquality.
SystemaLic
reviews musttake
accountof
studyquality
if
they areto
produce
unbiased estimates of
the
effects oftreatment.
However,
methodsfor
assessingtrial
quality
have
not
yet
been
futly
validated3i)35), se we cannotyet
be
sure that mechanismsfer
rating studyquality
are
truly
abgeto
discriminate
between
trials
that
are and are not
likely
to
be
biased.
How
potentially
biased
does
a studyhave
te
be
before
it
should nolonger
be
usedfer
clinicalsien-making?
The
answer shoulddepend
onthe
degree
of confidencethat
is
held
in
other
tion
that
pertains
to
the
clinicalquestion
athand,
As
a workingprineiple,
the
threshold
ofquality
should
be
that
the
study mustbe
ableto
provide
more certainty
than
the
reader alreadyhas.
Our
opinion
is
that,
in
practice,
there
will usuallybe
tle
point
in
reading clinicaltrials
that
do
not employ
true
randomisation,have
reasonablelevels
ofup, and
blind
assessors.
3.
Synthesis
offindings.
Idealty,
systematic reviews areaccompanied
by
meta-analysesthat
provide
pooled
estimates of treatment effects.
However,
sis
is
enlyindicated
whenthe
individual
studies arePhysical
Therapy
435
of sufficient
quality
and whenthere
is
sufficient
homogeneity
ofinterventions,
outcomes andfindings
acress studies.
When
heterogeneity
precludes
analysis, some authors cenduct "Ievels of evidence"
syntheses
in
whichthe
quality
of evidenceporting
a conclusionis
rated accordingto
a
termined
scale of studyquality
and consistency of
findings.
Unfortunately
the
findings
eflevels
ofdence
syntheses maydepend
heavily
onthe
ratingsystem used, and may
be
unduly sensitivete
the
findings
ofindividual
studies36),
These
three
problems
(publication
bias,
andthe
culties
in
assessingtrial
quality
and synthesisingtrial
findings)
lirrLit
our confidencein
the
cenclusions of
domized
trials
and systematic reviews.Nonetheless,
they
sheuld
be
put
in
perspective:
other sources of
tion
aboutthe
effects oftherapies
such as clinicalrience er
theory
or unconzrolled studies areprobably
prone
to
more seriousbias.
Mttny
readers are unabteto
discriminate
between,
studieswhich a,re
probabty
vatid andthose
tvhich areprobabty
not
Almost
all methodological surveys and mostatic reviews
in
physical
therapy
have
decried
the
ity
ofpublished
research(eg,
reference37).
This
is
tially
problematic
because
manyphysical
therapists
do
not
have
sufficienttraining
in
research methodolegyto
confidentiy
distinguish
between
studies ofhigh
andlow
quality.
There
is
a risk of many readersbeing
mis]ead
by
potentially
biased
studies.
The
evenzual so}utien mustbe
thatphysical
therapistswill
deveLop
the
skillsto
critically appraise cliniealresearch.
Most
undergraduate curricula nowteach
research methods and
increasingly
more explicitlyteach
critical appraisal of clinical research, so
in
the
nearfuture
we may
be
ableto
expect newgraduates
to
have
basic
criticat appraisal skills.
Those
whohave
alreadyated will
have
to seek outtraining
in
skills of criticalappraisaHf
they
areto
practise
evidence-basedphysical
therapy.
It
is
to
be
hoped
that
they
do
so withthe
sameenthusiasm
that
mostphysical
therapists
applyto
the
developmenl/
of other clinical skills.
Some
simple strategies may enhancephysical
pists'
abilitiesto
distinguish
between
high
andLow
ity
evidence.These
include
the use of methodologicalters"
4)3S)or use of methodological ratings
from
the
PEDro
database
to
screen outlow
quality
research.Secondary
sources of
publicatien,
such asthose
referredto
earlier,can
perferm
much ofthe
work of critical appraisalfer
clinicians who
lack
critical appraisal skills.Some
ofthese,
436
ge"utideV
sueh as
Cochrane
Systematic
Reviews,
arequite
uni-formly
ofhigh
qualtty
and cangenerally
be
censideredto
provide
an unbiased synthesis ofthe
Literature,
Onty
patient-centpmed
research can reatlytete
us aboutpeoples'
experiences
We
want clinicattrials
to
tell
us abouthow
much atherapy
affects apatient
in
terms
that
matterto
patients.
A
problem
with clinicaltrials
is
that
they
only measure outcomesthat
the
experimenterperceives
asimportant
andthey
do
notpermit
complete expression of whatpatients
feel
whengiven
aparticular
therapyS)i8)39),
At
enelevel
manytrials
do
measurethe
effects oftherapy
in
terms
that
patients
themselves
deem
to
be
important
Many
trials
now measure eutcomes such as"global
perceived
effect" or"preference
for
treatment''
because
it
is
thought
that
measurement ofthese
out-comesgives
patients
the
opportunityto
assignappre-priate
weighting to theirfeelings
oftheir
respenses totherapy.
Nonetheless,
these
single-dimensional outcomesprovide
little
opportunityfor
patients
to
expressthe
breadth
oftheir
feelings
about the effects oftherapies40),
The
needfor
patient-centered
outcomesin
clinicaltri-als suggests one
important
way(but
notthe
only way)in
whichqualitative
andquantitative
research cancom-plement
each otherin
evidence-basedpractiee.
Qualitative
research caninform
the
designers
of clinicaltrials
about what consumers see asthe
important
issues
when choosingtherapiesiS).
Such
censiderationsprobably
shouldbe,
but
rarely are,paramount.
thidence-based
p7iu]tice
removesthe
clinicatdecision-making role
from
ctinicians andgives
it
to
managersThere
is
a viewthat
evidence-basedpractice
takes
clin-ical
decision-making
out of clinicians'hands,
In
our view,this
is
notintrinsically
wrong:there
is
nointrinsie
right oftherapists
to
be
solely responsiblefor
clinical decision-making.Instead,
the
justification
for
clinician-as-decision-makerlies
in
the
reasonable expectatienthat
this
pro-vides
the
best
possible
care alld outcomes,Nonetheless,
Sackett
and colleagueshave
arguedthat
evidence-based
practice
does
not subjugateresponsibili-ty
for
clinicaldecision-making4i}.
Instead,
it
emphasizesthe
role of cliniciansin
using evidenceto
answer
their
own clinieal
preblems,
and removesthe
constraint oftra-dition
from
clinicaipractice,
In
evidence-based
practice
the
responsibilityfor
clinicaldecisiens
is
taken
awayfrom
how-to
textbooks
anddevolved
to
indiviclual
practition-ers and
their
patients.
ca30tsca8
£
What
Does
the
Evidence
Say?
The
massive recentgrowth
in
the
number ofran-domized
trials
and systematic reviewshas
provided
us with a wealth ofinformation
to
guide
clinicalpractice.
In
particular,
there
is
now muchinformation
aboutthe
effects of
physicai
therapy
interventions.
In
this
sectien webriefly
summarisefindings
from
ran-domized
trials and systematic reviews ofthe
effective-ness of a range ofphysical
therapy
interventions,
The
best
evidenee confirmsthe
value of some currentphys-ical
therapy
practices
andthe
ineffectiveness
of others.Some
little-used
interventions
have
been
found
to
pro-duce
remarkablybenefieial
effects.Chronic
muscutosketetatpain
Musculoskeletal
disorders
arethe
most common cause of chronicincapacity
in
industrialized
countries42).
Fortunately,
some ofthe
mostimpertant
advancesin
physical
therapy
have
been
in
the
managementof
these
problems,
parLicularly
chronicIow
back
pain,
osteoarthri-tis
ofthe
hip
andknee,
and rheumateid arthritis.
There
is
strong evidencefrom
recent studiesthat
sim-ple
interventions
provided
soon after symptom onset canprevent
the
development
of chronicback
pain.
For
exam-ple,
the
provision
of reassurance aboutthe
self-limiting natureof
mestlow
back
pain
and adviceto
returnto
normal activity
as
soon
aspossible
increases
the
rate ef returnto
workfor
workers withlew
back
pain43}.
On
the
etherhand,
encouraging restis
probably
harmfu143)").
Traditionally
the
mainstays efphysical
therapy
man-agement of musculoskeletalpain
have
been
massage, manualtherapy
{that
is,
manipulation andjoint
mobi-lization),
electrotherapy(such
astherapeutlc
ultrasound,shortwave
diathermy
andlow-energy
Iaser}
andthera-peutic
exercise.Current
evidence
paints
a
mixedpicture
ofthe
effects efthese
interventions.
There
has
been
lit-tle
rigorous researchinto
the effects of massage45}, sethe
clinicaibenefits
of massage,if
any, remain unsub-stantiated46),Manual
therapy
is
more effectivethan
place-bo
in
relievinglow
back
pain
but
it
is
notyet
clearif
it
is
more effectivethan
otherphysical
therapy
treat-ments26).
Most
electrotherapiesprobably
have
little
morethan
placebo
effects47}.The
mostpositive
findings
comefrom
recent studies oftherapeutic
exercise.Many
trials
and
several reviewshave
now shownthat
exercise canproduce
clinically worthwhile reductionsin
the
disabili-ty
andhandicap
associated with chroniclow
back
pain,
osteoarthritis of
the
hip
andknee,
and rheumatoidarthri-tis26)4S)4g).
effec-NII-Electronic Library Service
Evidence-Based
tive
in
contemporary clinicaltrials
differ
from
tradition-al exercise
in
twe
important
ways,First
contemporarypractice
is
to
consider
fermal
exercisepregrams
aspart
of a moreglobal
process
of `activityprescription'
involv-ing
astructured returnto
normalhome,
werk and social activities.Second,
wherepreviously
the
biological
sci-ences
provided
an exclusivetheoretical
basis
for
exer-ciseprescriptien,
contemporaryprograms
now alsedraw
from
behavioral
sciences.Many
exerciseprograms
now explicitlyincorporate
principles
of cognitive-behavioraltherapysc).
With
this
approaeh,patients
aretaught
to
exercise to
quotas
ratherthan
as symptomspermit,
spe-cific rewards
are
provided
when exercisequotas
eractiv-ity
goals
are met, andpain
behaviors
are not rewardedby
therapist
attention.Stress
urina7:yincontinence
About
25%
of women experienceinvoluntary
loss
of
urine,
and
abeut2.5%
reportthis
causes muchbether
or agreat
problem5i).
The
problem
of urinaryincontinence
in
women,including
genuine
stress urinaryincontinence,
can
be
treated
effectively withpelvic
floor
muscletrain-ing52).
Pelvic
floor
muscletraining
with weighted vaginalcones
(weights
inserted
into
the
vagina}substantially
increases
probability
ef cure orimprovement
compared
to
ne-exercise control conditions53).An
important
recenttrial
of
the
effects ofpelvic
floor
muscletraining
for
women withgenuine
stress urinaryincontinence
has
shewnthat
six months oftraining
with8
near-maximalpelvic
floor
muscle contractionsthrice
daily
produces
large
reductionsin
the
risk ofincontinence-related
prob-Iems
withsocial
life,
sexlife
andphysical
activity54).In
this
sample<mean
duration
of symptoms10
years),
absolute reductions
in
risk of each ofthese
problems
exceeded
35%,
implying
that
atleast
onein
three
womenexperiences
each
ofthese
benefits
from
exercise,It
is
not clearif
training
with weighted cones,biofeedback
(electromyographic
feedback
ofpelvic
floor
muscleactiv-ity)
or electrical stimulationproduces
better
outcomes comparedto
pelvic
fioor
muscletraining
alone52)54}(but
see reference55}.
nfovement
dyojltnction
resulting,1hom
strokeIn
terms
ofdisability-adjusted
life
years,
stroke ranks asthe
sixthhighest
cause efburden
ofdisease
world-wicle and
is
the
single mostimportant
cause oE severedisability
in
people
living
in
their
ownhomes56}.
Multidisciplinary
rehabilitationprograms
reducethe
odds ofdeath
orinstitutionalized
care andthe
odds ofdeath
or
dependency
after stroke(odcls
ratios ofO.80
andO.78),
probably
partly
by
reducingphysical
disability57),
There
Physical
Therapy
437
is
seme evidencethat
physical
training
reducesty,
andthat
moretraining
produces
better
eutcomes58}59).Seme
specific methods oftraining
showpromise,
altheughthe
extent ofthe
evidence relatingto
these
methedsis
Iimited
to
afew
high
quality
trials.
One
simple strategyis
to
constrainthe
unaffectedupper
Iimb
to
''force" useof
the
affected upperlimb60).
Another
effective strategy
is
to
suspendpatients
In
aharness
above atreadmilt
to
enable
practice
of walking withpartial
body
weightPort6T).
Acute
and ciironic respiratorydisectse
The
role ofphysieal
therapy
in
prevention
andagement of
pulmonary
disease
has
been
debated
fer
decades,
Publication
of severalimportant
systematicreviews and clinical
trials
has
clarifiedthe
situation what.Pre-
andpost-surgical
prophytactic
chest
physical
apy reduces morbidity
fellewing
major abdominalsurgery62)63),
For
example,Olsen
and colleagues showedthat
prophylactic
chestphysical
therapy
reducedthe
dence
ofpulmonary
complications after major abdominalsurgery
from
27%
to
6%63).
This
implies
that
onage, one
pulmonary
complieationis
prevented
for
everyfive
patients
treated,
It
is
notyet
clear which specificinterventions
are most effective,Prephylactic
chesticat
therapy
has
li[tle
effect when reutinely administeredfollowing
coronary arterybypass
surgery or minorabdominal surgery, or
during
the
intubation
period
lewing
routine cardiac surgery(eg,
referenee64),
This
is
consistent with
the
position
that
prophylactic
ehestical
therapy
is
of mostbenefit
to
patients
atthe
highest
risk of
post-operative
eomplications63).Pulmonary
rehabilitationprograms
typically
invelve
upper and
tower
body
exercise(usually
treadmill
ing
or stationary cycling), and mayinclude
ventilatory'
muscletraining,
counseling and education.A
recenttematic
review cenfirmsthe
viewthat
pulmonary
bilitation
pregrams
canincrease
walkingdistance
andhealth-related
quality
oflife
in
peeple
with asthma andchronic obstructive
puimonary
disease65).
A
recenttrial
indicates
that
rehabilitation mayalse
redueeduration
ofhospitalisation,
but
not number of medical consultations66).
An
important
and relatively newintervention
is
the
application ef nocturnat ventilatory support to
patients
with sieep-disordered
breathing,
particularty
patients
with chronic obstructive
pulmenary
disease,
cular
diseases
orinjury,
er cysticfibrosis.
Nocturnal
assisted
ventilationproduces
large
reductionsin
one-yearmortality and
hypoventilation-related
symptoms67).43s
mp"utthl
Prevention
offo,tts
in
the
eldertyOne
in
three
elderpeople
fall
atleast
once ayear68),
There
is
strellg evidence that multifacetedinterventions
targeting
identified
riskfactors
reducefalls
riskin
olderpeopte69).
Well
designed
studies suggestthat
it
is
sary
to
prescribe
suchinterventions
for
about eightple
Lo
prevent
onefall
per
year
in a commun]ty
・
70)
taIlg
.
Knewn
riskfactors
for
falls
such as weakness oflower
limb
muscles andpoor
balance
arepotentially
modifiableby
exercise, sophysiotherapists
and othershave
vided exercise
programmcs
for
olderpeople
at risk offalling.
Several
trials
and reviews conc]udethat
exercisecan reduce risk ef
falLs69)7t).
One
individually
tailored
home
based
strength andbalance
training
prescribed
by
trained
heaith
professionals
has
been
found
effectivein
four
trials72).
Summary
and
Conclusions
Evidence-based
practice
involves
the systematie andcritical use of
high
quality
c[inical researchfor
clinical
decision
making.The
increase
hi
the
volume andty
of clinical researchin
physical
therapy
has
meantthat
there
is
now eneugh researchto
support clinicalsien-making.
Several
objectionshave
been
raised againstbased
practice.
Most
of these ebjectionshave
somebasis
in
fact.
Nonetheless,
the
legitimate
objectionsto
based
practice
have
tobe
weighed againstthe
obviousbenefit
ofbasing
clinicaldecision-making
enhigh
ty
research.There
are now many randomizedtrials
supportingthe
effectiveness of
key
physiotherapy
interventions,
ularly
those
interventions
involving
exercise,These
als
provide
the
researchinfrastructure
that
justifies
theprofession
ofphysical
therapy.
References
1}
Herbert
RD,
MaherCG
MoseLey
AM,
Sherrington
C/
Regular review/ effective physiutherapy.ew
323,
788-790,
2001.
2)
Hcrbert
RD,Sherri"gton
C,
Maher
C,
Mosetey
AM,
based
practicc-imperfecL
but
necessary. Physi,othetmpy 7-,eo7yand
Pi'actice
17,201-21],2001.
3)
Moselev
AM,Herbert
RD,
Sherrington
C,
Maher
CG'.
Evidencefor physioLherapy practice/ a survey of
the
l'hysiotherapyEvidence
Database
(PEDro),
Aitstral'ianJburnat qf'Ph/usiotherapy48/43-49,
2002.
4)
Sa[ketL
DL,
Struus
SE,
RichurdsonWS.
Rose"berg
W,
I-IaynesRB/
Evidence-Based
Medicine/How
toPractice
andTeach
EBM
(2nd
ed).Edinburgh/
Churchill
Livingstone,
2000.
5}
Carr
JH,
MungovanSF,
Shepherd
RB,
Dean
CM,
NordholmI.A,
Physlcal
thcrapyin
stroke rehabilitation/ basesfor
Australian
physicul therapists' choice of treatment. Physiothermpu
TheoTz/
and IZractice10/
201-209,
i994.
6)
Nilsson
LM,
Nordholm
LA/ Physical therupyin
stroketation/
bases
forSwedish
physical therapists' choice oftreatmeiit.
I'hysiotherap/u IVteotyand 1'ractice
8/
49-55,
1992.ig3ogegse
7)
8}
9)
10)
lv
12}
i3)
14)
15>
16)
17)18)
19)
20)2D
22)
23)2,DZ5)
26) 27)28)29)30)
3D
Nationnl
I'IealthandMedical
Research
Council/
How
to
Use
the
Evidence/
Asscssment
undApplictnion
ofScientthc
Evidence.Canbcrra/
Bietext,
2000.
GreenhaLgh
T/ rS'arrativebased
medicine/ nurrativebascd
medi-ci"e inan evidence
based
world.BnV
318,
323-325,
1993.C]haLmers
TC,
Celane
P,
Sacks
HS,
Smith
H/
Blas
in treatment ussignmentin
ctontroUed cainical trinls.Nbtv
England
Jou'tTnatqfMed'cine
309/
1358-l361, 19S3.Coiditz
GA,
MlllerJN,
Mostelter
F,
How
studydesign
affects out-comcs in coniparisons of therapy-1/medical,Statistics
i,n
Medi
7/neS/
441-454, 1989.LiiTde
K,
Scholz
)JI,
Ramirez
G,
Cluusius
pa,
Melchart
D,Jonas
WB/
Impact
of study quality on outcomein
p]acebo-conLrolled trlMls of hoineepat.hy. Joumatof
Ctinicat
opidemiology
52/
631-636.
1999.Ivliller
JN,
Coiditz
GA,
Mosteller
F/
Hew
tLdydesign
affects ouLconies
in
comparTsons of therapy-II/ surgical.Stat-lst'ies
i'itMbdici.ne
S,455-466, l989.Sacks
H,Chalmers
TC,
Smith
H/
Randomized
versushistericHl
cont.rols
for
clinieal triHls.nmeiiean
-iovtnytat
Qf
dedicine
72/233-24e.
1982.
Benson
K.
Hartz AJ/A
comparisen of observarional studies and randomized. ccmtrollecltrials.New EngtandJournat
of
Med'icine
342/
IS78-l886,
2000.Concar[]
J,
ShahN.
Horwitz
RL
Randomized,
conLrolled trials.observationak studies,and
the
hierarchy
ef researchdesigns.
Neu.,
Englaud
Joumat
of
Medicine 342/ ]887-1892,200e.
Lijmer
JG.
Mol
BW, HeisterkampS,
Bonsel
GJ,
Prins
MH,
vander
Meuien
JH,
BossuytPM/
EmpirieaL
evidencc ofdesTgn-relat-ed
bias
in
studies efdiagnostic
Lests..L4rm
282,
1061-1066,
]999.de
Bie
R/
Critical
appraisul ofprognestic studies/ anintroduction.
IVtysiotherapy
7heory and Practice17,
161-l71,
2eOl.
Ritchie
JE/
Using qualitativeresearch to e"hance theevldence-based
practice
ofhealth
care providers,Azastrcstian
Jbutveal
qfPt,ysiothermpy
4S,251-256, 1999.Ritchie
JE/
Cuse
series research/ a casefor
qualitativemethodin
assembling evidence. PhysiotheTapy7heo?v
andPtuctice
17/127-135, 2001.
Di
Fabio
RP/
Mvth
of evidence-based practice.Jbutm.at
qt'O,V,opaedic
a,ndhb]orts
Phys'tcaZ 77ie,npy29/
632-633,
1999.
Feinstein
AR, I{orwitzRI/
Prob]ems
in
the"evidence"
of"evi-clence-based
medicine". America",fournal
of
Med'if/ine
103/
529-535,
1997.Evidence-Based
Care
ResourceGroup/
Evidence-based
care-l.Setting
priorities/
howimportunt
is
the problem?mmJ
150/
]249-1254.
I994.Turner P,
Whitfield
TW/
Physiotherapists'
use ofevidence busedpractice/
a c/r{]ss-national study.Physiothe7'apy
Researchintennationat
2:l7-29, 1997.van
der
Valk RWA.Dekker
J,
vanBaar
ME/
Physicai
therapyfor
puLients withbnck
pain.Physiothe,upy
Sl/
345-351,
l995.Maher
C,
Latimer
J,
Refshauge
K/
Prescription
of attivityfor
lovv
back
pnin/what works? AustralianJbur:?,al
efPhysiotherapy 45/121-132, 1999,van Tutdcr
MW,
Koes
BW,
Bouter
LM/
Conservative
treatmentof acute and chronlc nonspecific
low
back
puin.A
systemativreview of randomfzed controlted trials of the most cemmon
inLer-ventions.
spine
22,
2128-2156,
l997.
Haynes
B.
IIaines
A/
Getting
researchfindings
intopractice/bar-riers and
bridges
to evidencebased
clinical practict.'.ilnti317/
273-276,
1998.
Walkcr-Dilks
C/Searching
the physiotherapy evidence-basedLit-erature.
Ph,ysiotheTupy
Iheo7v andl-rtact.ice
17/137-142. 200].Bithell
C/Evidence-based physiotherapy/ some thoughts on 'bestevidence'. I]hysiotlteropy