originally published online september 12, 2013 - … directaccess ptj 1-14.pdforiginally published...

doi: 10.2522/ptj.20130096Originally published online September 12, 2013

2014; 94:14-30.PHYS THER. Heidi A. Ojha, Rachel S. Snyder and Todd E. DavenportTherapy Episodes of Care: A Systematic ReviewDirect Access Compared With Referred Physical

http://ptjournal.apta.org/content/94/1/14found online at: The online version of this article, along with updated information and services, can be

Collections

Systematic Reviews/Meta-analyses Policies, Positions, and Standards

Health Services Research Health Care System

Economics Direct Access

in the following collection(s): This article, along with others on similar topics, appears

e-Letters

"Responses" in the online version of this article. "Submit a response" in the right-hand menu under

or click onhere To submit an e-Letter on this article, click

E-mail alerts to receive free e-mail alerts hereSign up

at APTA Member on March 2, 2014http://ptjournal.apta.org/Downloaded from at APTA Member on March 2, 2014http://ptjournal.apta.org/Downloaded from

http://http://ptjournal.apta.org/content/94/1/14

http://ptjournal.apta.org/cgi/collection/direct_access

http://ptjournal.apta.org/cgi/collection/direct_access

http://ptjournal.apta.org/cgi/collection/economics

http://ptjournal.apta.org/cgi/collection/economics

http://ptjournal.apta.org/cgi/collection/health_care_system

http://ptjournal.apta.org/cgi/collection/health_care_system

http://ptjournal.apta.org/cgi/collection/health_services_research

http://ptjournal.apta.org/cgi/collection/health_services_research

http://ptjournal.apta.org/cgi/collection/policies_positions_and_standards

http://ptjournal.apta.org/cgi/collection/policies_positions_and_standards

http://ptjournal.apta.org/cgi/collection/systematic_reviews

http://ptjournal.apta.org/cgi/collection/systematic_reviews

http://ptjournal.apta.org/letters/submit/ptjournal;94/1/14

http://ptjournal.apta.org/letters/submit/ptjournal;94/1/14

http://ptjournal.apta.org/subscriptions/etoc.xhtml

http://ptjournal.apta.org/subscriptions/etoc.xhtml

http://ptjournal.apta.org/




Direct Access Compared WithReferred Physical Therapy Episodesof Care: A Systematic ReviewHeidi A. Ojha, Rachel S. Snyder, Todd E. Davenport

Background. Evidence suggests that physical therapy through direct access mayhelp decrease costs and improve patient outcomes compared with physical therapyby physician referral.

Purpose. The purpose of this study was to conduct a systematic review of theliterature on patients with musculoskeletal injuries and compare health care costsand patient outcomes in episodes of physical therapy by direct access compared withreferred physical therapy.

Data Sources. Ovid MEDLINE, CINAHL (EBSCO), Web of Science, and PEDrowere searched using terms related to physical therapy and direct access. Includedarticles were hand searched for additional references.

Study Selection. Included studies compared data from physical therapy bydirect access with physical therapy by physician referral, studying cost, outcomes, orharm. The studies were appraised using the Centre for Evidence-Based Medicine(CEBM) levels of evidence criteria and assigned a methodological score.

Data Extraction. Of the 1,501 articles that were screened, 8 articles at levels 3 to4 on the CEBM scale were included. There were statistically significant and clinicallymeaningful findings across studies that satisfaction and outcomes were superior, andnumbers of physical therapy visits, imaging ordered, medications prescribed, andadditional non–physical therapy appointments were less in cohorts receiving phys-ical therapy by direct access compared with referred episodes of care. There was noevidence for harm.

Data Synthesis. There is evidence across level 3 and 4 studies (grade B to CCEBM level of recommendation) that physical therapy by direct access comparedwith referred episodes of care is associated with improved patient outcomes anddecreased costs.

Limitations. Primary limitations were lack of group randomization, potential forselection bias, and limited generalizability.

Conclusions. Physical therapy by way of direct access may contain health carecosts and promote high-quality health care. Third-party payers should considerpaying for physical therapy by direct access to decrease health care costs andincentivize optimal patient outcomes.

H.A. Ojha, PT, DPT, OCS,FAAOMPT, Department of Physi-cal Therapy, College of HealthProfessions and Social Work, Tem-ple University, 3307 N Broad St,Philadelphia, PA 19140 (USA).Address all correspondence to DrOjha at: [email protected].

R.S. Snyder, PT, DPT, OCS, StLuke’s Physical Therapy, Easton,Pennsylvania.

T.E. Davenport, PT, DPT, OCS,Department of Physical Therapy,Thomas J. Long School of Phar-macy and Health Sciences, Univer-sity of the Pacific, Stockton,California.

[Ojha HA, Snyder RS, DavenportTE. Direct access compared withreferred physical therapy episodesof care: a systematic review. PhysTher. 2014;94:14–30.]

© 2014 American Physical TherapyAssociation

Published Ahead of Print:September 12, 2013

Accepted: August 30, 2013Submitted: March 11, 2013

Research Report

Post a Rapid Response tothis article at:ptjournal.apta.org

14 f Physical Therapy Volume 94 Number 1 January 2014 at APTA Member on March 2, 2014http://ptjournal.apta.org/Downloaded from



An estimated 53.9 million peo-ple in the United States reporthaving 1 or more musculo-

skeletal disorders, with per capitamedical expenditures averaging morethan $3,578.1 As musculoskeletalconditions represent some of theleading causes of restricted activitydays,2 many of these individuals seekcare from or are referred to a physi-cal therapist. In 2007, Americans ofall ages had more than 164 millionambulatory visits for physicaltherapy.3

Individuals typically seek physicaltherapy services through eitherdirect access or physician referral.The term direct access, in thisreport, will be defined as patientsseeking physical therapy caredirectly without first seeing a physi-cian or physician assistant to receivea script or referral for physical ther-apy services. Some studies have sug-gested that early or direct access tophysical therapy can reduce waitingtime, improve convenience, reducecosts for the patient and healthcare system, and improve recoverytime.4–6 The results of these studiesdirectly support recent health carereform efforts in which legislatorsand health care providers havesought to provide efficient carethrough cost reduction and optimiz-ing patient outcomes.

In response to the growing literaturesupporting physical therapy’s role inprimary care, 47 out of 50 states(United States) currently have legis-lation that provides for some formof direct access to physical therapy.However, various barriers preventphysical therapists from practicingin a direct access capacity. Many ofthese 47 states limit direct access tocertain physical therapists’ qualifica-tions, specialty areas, or condition/diagnostic codes. Furthermore, phys-ical therapists may require referralsfrom medical providers due to legalconstraints, third-party payer require-

ments for reimbursement, and hos-pital bylaws. Finally, there is a lack ofpublic awareness and autonomoushealth-seeking behavior among con-sumers.7 Consequently, even thoughmost physical therapists have directaccess privileges through their statepractice acts, the large majority ofpatients are still managed throughepisodes of care that are initiated byphysician referral.

Despite the growing body of scien-tific literature in support of con-sumer direct access to physical ther-apy, the only systematic review that,in part, evaluated the impact of phy-sician referral versus direct access onoutcomes and costs was publishedin 1997 by Robert and Stevens.4 Thereview4 found that the main advan-tages for direct referral to physicaltherapy were significant reductionsin waiting times, convenience, andreduced costs for the patient. How-ever, there was little evidence in thepublished literature at that time tomake conclusions about recoverytime, outcomes, or cost to the healthsystem. There have been a numberof articles published since the mid-1990’s on this topic,8–15 and we areunaware of any recent reviews pub-lished on this topic.

The purpose of this study was toestablish the effects of direct accessand physician-referred episodes ofcare in individuals receiving physicaltherapy based on a systematicreview of peer-reviewed literature.Published data regarding clinicaloutcomes, practice patterns, utiliza-tion, and economic data were usedto characterize the effects of directaccess versus physician-referred epi-sodes of care. We hypothesized thatpolicies permitting patients to seekphysical therapy directly wouldresult in decreased health care costsand similar patient outcomes. Wealso hypothesized that there wouldbe no evidence of increased harmrelated to direct access compared

with physician-referred episodes ofphysical therapy. The relevance of asystematic review at this time is thatadditional scientific weight can beprovided to guide the physical ther-apist’s role in health care reformand serve as a concise report toimprove the ability of consumers,legislators, hospital administrators,and third-party payers to synthesizethe existing literature and make con-clusions regarding the quality andcost-effectiveness of primary accessphysical therapy.

MethodData Sources and SearchesDatabases of CINAHL (EBSCO)(restricted to humans, January1990–July 2013), Web of Science(restricted to articles, 1990 andlater), and PEDro (1990 and later)were searched last on July 5, 2013.Limits were not placed on languagewhen conducting all searchesbecause we did not want to excludearticles written in the Spanish lan-guage, one author’s second lan-guage. All searches were restrictedto 1990 to present because wewanted to specifically focus onmore recently published literatureto improve generalizability of results,reflecting changes in modern prac-tice patterns and updated interpre-tations of the search terms “directaccess” and “open access.” Wesearched the databases using combi-nations of the key words “directaccess,” “primary care,” “physicaltherapy,” “physiotherapy,” and “openaccess.” In addition to these keywords, we searched Ovid MEDLINE(1990 and later) using a comprehen-sive list of Medical Subject Headings(MeSH) terms related to our topic.The full electronic search strategyand results for the Ovid MEDLINEdatabase are listed in Table 1 as anexample of the searches performedin this review.

Direct Access Versus Referred Physical Therapy Episodes of Care

January 2014 Volume 94 Number 1 Physical Therapy f 15 at APTA Member on March 2, 2014http://ptjournal.apta.org/Downloaded from



Study SelectionStudies had to satisfy all of the fol-lowing criteria to be included inthis review: (1) included patientswith greater than 85% musculoskel-etal injuries treated by a physicaltherapist in an outpatient setting,(2) included original quantitativedata of at least one group thatreceived physical therapy throughdirect access or direct allocation to aphysical therapist without seeing aphysician, (3) included originalquantitative data for at least onegroup that received physical therapythrough physician referral, (4) greaterthan 50% of the patients in bothgroups had to have received physicaltherapy, and (5) included assessmentof at least one of the following: out-comes of physical therapy (improve-ment or harm), cost, or outcomemeasures that would affect cost oroutcomes (use of imaging, pharma-cological interventions, consultantappointments, and patient satisfac-tion). Among all articles read in fulltext, studies were excluded if theywere written in a language that theauthors did not speak (all languagesexcept English and Spanish).

Data ExtractionTitles and abstracts were screenedby the authors (H.A.O. and T.E.D.).Full texts were obtained for any arti-cle that could not be ruled out basedon the specified inclusion criteria.The Figure lists our search strategy,also referenced in the “Results” sec-tion of the article. The primary char-acteristics were extracted from eachstudy. Results of the direct accessand physician referral groups fromeach study were extracted for out-comes of interest at all time frames(most studies collected outcomesduring the course of physical ther-apy, at discharge, or both). Theseoutcomes of interest were: cost-effectiveness, number of physicaltherapy visits, discharge outcomes,imaging use, medication use, patientor physician satisfaction, number ofadditional referrals, additional caresought, or evidence of harm to thepatient, physical therapist, or facility.Because of the conceptual heteroge-neity in dependent variable measure-ments and lack of reports of variabil-ity around point estimates, we wereunable to pool data and calculateeffect sizes. Therefore, means or dif-ferences between means were listedfor each outcome measure

extracted, and standard deviationsand ranges were reported as avail-able (if not reported, the study didnot report the information).

Quality AssessmentThe validity of studies using abetween-group comparison wasevaluated by 2 authors not blindedto authors or journals. We used theOxford 2011 Centre of Evidence-Based Medicine (CEBM) recommen-dations to rate each article’s level ofevidence16 and the Downs and Blackchecklist17 to assign a methodologi-cal quality score to each articlebecause all of our included studieswere nonrandomized. See Table 2and Appendix 1 for descriptionsof the CEBM levels of evidence andDowns and Black checklist criteria.The level of evidence was deter-mined across studies for the purposeof assigning a grade of recommenda-tion to synthesize results; however,level of evidence was not utilized asan exclusion criterion due to the lim-ited number of articles that met ourdefined criteria.

All studies were independentlyscored using a modification of theDowns and Black tool17 by 2 review-ers (H.A.O. and R.S.S.) who wereblinded to each other’s results. Afterscoring, any disagreements wereresolved by discussion (T.E.D.). TheDowns and Black checklist is a toolthat can be used to assess the meth-odological quality of nonrandomizedstudies. We developed guidelines,specific to our study type, toimprove agreement between raters(Appendix 1). Reliability betweenreviewers’ initial Downs and Blackchecklist scores was calculated usingthe kappa coefficient. Similar toother previously publishedreviews,18–20 the tool was slightlymodified for use in our study bydropping 2 checklist items from ouranalysis. We decided that criterion17 (“In trials and cohort studies, dothe analyses adjust for different

Table 1.Examples of Search Words and Medical Subject Headings (MeSH) Terms Used inOvid MEDLINE

Search Word/MeSH Term (1990 to Present)No. of Articles

Located

Physical Therapy Modalities AND Family Practice 120

Physical Therapy Modalities AND Referral and Consultation 200

Physical Therapy Modalities AND Musculoskeletal Diseases 315

Professional Competence AND Physical Therapy Specialty 66

Community Health Centers (Organization andAdministration) AND Referral and Consultation

103

Family Practice AND Physical Therapy Department, Hospital 2

Outpatient Clinics, Hospital (Utilization) AND Referral andConsultation

190

Physical Therapy Modalities AND Delivery of Health Care 77

Physical Therapy Modalities AND Primary Health Care 150

Total no. of articles located in database 1,223





Figure.The flowchart outlines the search strategy. Because of the conceptual heterogeneity in dependent variable measurements and lackof reports of variability around point estimates, we were unable to pool data and calculate effect sizes.





Tab

le2.

Stud

yC

hara

cter

istic

san

dRe

sults

ofIn

clud

edSt

udie

sa

Stu

dy

and

Co

un

try

Qu

alit

yR

atin

gSa

mp

leSi

zeG

rou

ps,

Stat

isti

csb

Co

st

Ph

ysic

alT

her

apy

Vis

itsc

Imag

ing

Ord

ered

Ph

arm

aco

log

ical

Tre

atm

ent

Ap

po

intm

ents

Pat

ien

tSa

tisf

acti

on

DC

Ou

tco

mes

or

Har

m

Hac

kett

etal

,15

1993

Uni

ted

King

dom

CEB

Md

3D

&B

13/2

6Tw

oru

ralp

ract

ices

,�

42%

spin

alin

jurie

s,th

ere

stex

trem

ityin

jurie

s(�

95%

msk

)In

sura

nce:

NH

SI:

n�85

C:

n�13

3

I:op

enac

cess

Mea

nco

stto

pts

�£9

.55

($14

.83)

X�

7.2

5.1%

refe

rred

for

imag

ing

48%

rece

ived

pre

scrip

tion

med

icat

ions

2.3%

had

GP

cons

ulta

tions

,2.

5%re

ferr

edto

spec

ialis

ts

93.2

%sa

tisfie

dfD

ays

ofw

ork

mis

sed

due

todx

�10

.4

C:

refe

rred

byor

thop

edic

cons

ulta

nts

Mea

nco

stto

pts

�£4

7.94

($74

.48)

X�

7.0

13.6

%re

ferr

edfo

rim

agin

g84

%re

ceiv

edp

resc

riptio

nm

edic

atio

ns

1.5%

had

GP

cons

ulta

tions

,8.

2%re

ferr

edto

spec

ialis

ts

84.1

%sa

tisfie

dfD

ays

ofw

ork

mis

sed

due

todx

�27

.9

Stat

istic

sP�

.01

(95%

CI�

£12.

41,

£63.

65)

NS

PN

RP�

.001

eP

NR

P�.0

1eP

NR

Hol

dsw

orth

and

Web

ster

,12

2004

Scot

land

CEB

M3

D&

B16

/26

Pts

with

msk

inju

ries

from

26ge

nera

lp

ract

ices

,In

sura

nce:

NH

SI:

n�76

C:

n�26

4

I:se

lf-re

ferr

als

Few

erp

hysi

cal

ther

apy

visi

tsFe

wer

GP

cont

acts

3m

oaf

ter

phy

sica

lthe

rap

y

VAS

scor

ede

crea

sed

from

5.7

(SD

�2.

3)to

2.7

(SD

�1.

7)78

%ac

hiev

edgo

als

79%

com

ple

ted

thei

rco

urse

ofp

hysi

cal

ther

apy

C:

GP

refe

rral

grou

pg

Mor

ep

hysi

cal

ther

apy

visi

tsM

ore

GP

cont

acts

3m

oaf

ter

phy

sica

lth

erap

y

VAS

scor

ede

crea

sed

from

5.7

(SD

�2.

2)to

3.2

(SD

�1.

6)63

%ac

hiev

edgo

als

60%

com

ple

ted

thei

rco

urse

ofp

hysi

cal

ther

apy

Stat

istic

sP�

.038

P�.0

113

VAS

P�.0

11,

Goa

lsac

hiev

edN

S,P�

.079

Cou

rse

com

ple

ted

P�.0

04

Hol

dsw

orth

etal

,13

2007

Scot

land

CEB

M3

D&

B16

/26

Pts

with

msk

inju

ries

from

26ge

nera

lp

ract

ices

thro

ugho

utSc

otla

ndIn

sura

nce:

NH

SI:

n�64

8C

:n�

1,79

5

I:se

lf-re

ferr

als

n�64

8A

vera

geco

stp

erep

isod

eof

care

£66.

31($

136.

02)

X�

4.6

(ran

ge�

1–22

)Ra

diog

rap

hs�

4.00

32.2

%re

ceiv

edm

edic

atio

nsh

18%

visi

ted

GP

onm

ore

than

one

occa

sion

67%

achi

eved

goal

s

C:

GP

refe

rral

grou

pg

n�1,

795

Ave

rage

cost

per

epis

ode

ofca

re£8

8.99

($13

8.26

)

X�

4.6

(ran

ge�

1–22

)Ra

diog

rap

hs�

7.43

44.1

%re

ceiv

edm

edic

atio

nsh

39%

visi

ted

GP

onm

ore

than

one

occa

sion

65%

achi

eved

goal

s

Stat

istic

sN

RN

SP�

.01

P�.0

001

P�.0

01N

S,P�

.82

Web

ster

etal

,14

2008

Scot

land

CEB

M3

D&

B16

/26

Pts

with

msk

inju

ries

from

26ge

nera

lp

ract

ices

Insu

ranc

e:N

HS

I:n�

648

C:

n�1,

795

I:se

lf-re

ferr

als

17%

cont

actin

gG

P1

mo

afte

rD

CSa

tisfie

di �79

%

C:

GP

refe

rral

grou

p19

%co

ntac

ting

GP

1m

oaf

ter

DC

Satis

fiedi �

74%

Stat

istic

sN

S,P�

.219

P�.0

01

(Con

tinue

d)




Tab

le2.

Con

tinue

d

Stu

dy

and

Co

un

try

Qu

alit

yR

atin

gSa

mp

leSi

zeG

rou

ps,

Stat

isti

csb

Co

st

Ph

ysic

alT

her

apy

Vis

itsc

Imag

ing

Ord

ered

Ph

arm

aco

log

ical

Tre

atm

ent

Ap

po

intm

ents

Pat

ien

tSa

tisf

acti

on

DC

Ou

tco

mes

or

Har

m

Mitc

hell

and

deLi

ssov

oy,9

1997

Uni

ted

Stat

es

CEB

M3

D&

B18

/26

Acu

te/s

por

adic

msk

-re

late

ddi

sord

ers,

adul

tsag

ed�

65y

and

thei

rch

ildre

n,BC

BSI:

n�25

2C

:n�

353

I:di

rect

acce

ss$1

,004

pai

dcl

aim

sp

erep

isod

eof

care

jX

�20

.2Ra

diol

ogy

clai

ms

0.32

Dru

gcl

aim

s1.

47M

ean

hosp

ital

adm

its0.

25

C:

phy

sici

anre

ferr

al$2

,236

pai

dcl

aim

sp

erep

isod

eof

care

jX

�33

.6Ra

diol

ogy

clai

ms

1.02

Dru

gcl

aim

s3.

13M

ean

hosp

ital

adm

its0.

64

Stat

istic

sP�

.001

P�.0

001

P�.0

1P�

.01

P�.0

1

Leem

rijse

etal

,820

08Th

eN

ethe

rland

s

CEB

M4

D&

B22

/26

PTs

atp

rivat

ep

ract

ices

liste

din

ada

taba

se;

spec

ialis

tPT

s(ie

,p

eds)

excl

uded

I:n�

2,97

7C

:n�

7,53

3

I:di

rect

acce

ssX

�8.

1(S

D�

6.6)

Goa

lsfu

llyac

hiev

edat

DC

,i85

0(7

2.6%

)

C:

phy

sici

anre

ferr

alX

�10

.5(S

D�

8.9)

2,10

2(6

3.3%

)

Stat

istic

sP�

.001

P�.0

01

Moo

reet

al,1

020

05U

nite

dSt

ates

CEB

M4

D&

B13

/26

Mili

tary

pat

ient

sse

enby

PTs;

msk

dxI:

n�50

,799

C:

n�42

1,21

9

I:di

rect

acce

ssN

oha

rmk

C:

PTs

via

refe

rral

No

harm

k

Stat

istic

sP

NR

Pend

erga

stet

al,1

1

2012

Uni

ted

Stat

es

CEB

M4

D&

B22

/26

Adu

lts(1

8–64

y)tr

eate

din

outp

atie

ntcl

inic

s(p

rivat

eor

hosp

ital

base

d)on

priv

ate

insu

ranc

ep

lan

(BC

BS)

atle

ast

87%

msk

I:n�

17,4

97C

:n�

45,2

10

I:se

lf-re

ferr

edM

ean

allo

wab

leam

ount

s:PT

�$5

03.1

2(S

D�

$478

.18)

,no

n-PT

�$5

26.2

6(S

D�

$1,4

48.9

5)

X�

5.90

(SD

�5.

55)

C:

phy

sici

anre

ferr

alM

ean

allo

wab

leam

ount

s:PT

�$6

05.4

9(S

D�

$549

.61)

,no

n-PT

�$6

78.6

4(S

D�

$1,7

44.1

1)

X�

7.00

(SD

�6.

09)

Stat

istic

sBo

thP�

.001

P�.0

01

aI�

inte

rven

tion

grou

p,

C�

com

par

ison

grou

p,

D&

B�D

owns

and

Blac

kch

eckl

ist

(see

Ap

pen

dix

1fo

rcr

iteria

),N

HS�

Nat

iona

lHea

lthSe

rvic

e,BC

BS�

Blue

Cro

ssBl

ueSh

ield

,p

ts�

pat

ient

s,C

EBM

�C

entr

efo

rEv

iden

ce-

Base

dM

edic

ine,

dx�

diag

nosi

s,D

C�

disc

harg

e,PT

�p

hysi

calt

hera

pis

t,m

sk�

mus

culo

skel

etal

,p

eds�

ped

iatr

ics,

95%

CI�

95%

confi

denc

ein

terv

al,

GP�

gene

ralp

ract

ition

er,

NR�

not

rep

orte

d,N

S�no

tsi

gnifi

cant

.b

Stat

istic

aldi

ffere

nce

betw

een

Ian

dC

grou

ps.

cPh

ysic

alth

erap

yvi

sits

per

epis

ode

ofca

re(m

ean

acro

ssal

lpat

ient

s).

dLe

vels

ofev

iden

cear

eba

sed

onth

eO

xfor

d20

11C

EBM

leve

lsof

evid

ence

:le

vel1

�sy

stem

atic

revi

ewof

rand

omiz

edtr

ials

orn�

1tr

ial;

leve

l2�

rand

omiz

edtr

ialo

rob

serv

atio

nals

tudy

with

dram

atic

effe

ct;

leve

l3�

nonr

ando

miz

edco

ntro

lled

coho

rt/f

ollo

w-u

pst

udy;

leve

l4�

case

-ser

ies,

case

-con

trol

,or

hist

oric

ally

cont

rolle

dst

udie

s;le

vel5

�m

echa

nism

-bas

edre

ason

ing.

eP

valu

ein

clud

esp

ract

ice

A(p

hysi

cian

owne

d)�

pra

ctic

eB

(dire

ctac

cess

)vs

pra

ctic

eC

(ref

erra

l);95

%C

I�26

.1%

to44

.0%

.fEv

alua

ted

man

agem

ent

ofth

eir

cond

ition

asav

erag

eor

abov

e.g

GP-

sugg

este

dre

ferr

algr

oup

resu

ltsex

clud

ed.

hM

edic

atio

ns�

nons

tero

idal

anti-

infla

mm

ator

ydr

ugs

and

anal

gesi

cs.

iPe

rcen

tsa

tisfie

d�p

erce

ntsa

tisfie

dor

very

satis

fied.

jPa

idcl

aim

sfo

ral

lser

vice

s/dr

ugs

per

epis

ode

ofca

re.

kN

oad

vers

eev

ents

resu

lting

from

PTdx

orm

anag

emen

t,no

stat

elic

ense

sm

odifi

edor

revo

ked

for

disc

iplin

ary

actio

n,no

litig

atio

nca

ses

filed

agai

nst

US

gove

rnm

ent.




lengths of follow-up of patients, or,in case-control studies, is the timeperiod between the intervention andoutcome the same for cases and con-trols?”) was not a good evaluation ofquality because the follow-up periodin many studies was initial evaluationto discharge, which was influencedby one of our primary outcome mea-sures (number of physical therapyvisits). Criterion 27 (“Did the studyhave sufficient power to detect aclinically important effect where theprobability value for a differencebeing due to chance is less than5%?”) also was not scored becausewe consulted a statistician whobelieved that significance foundshould not be influenced by posthoc power analyses and a differencebetween groups is either significantor not at study end, regardless ofhow much power was assumed a

priori.21 The maximum score on thescale was 26, as one item had apotential of 2 points and we omitted2 criteria (Tab. 2).

Data Synthesis and AnalysisAfter assigning a level of evidence toeach article according to the CEBMcriteria, the data were synthesizedby the primary author (H.A.O.), strat-ified by outcome measure utilizinggrades of recommendation A to Daccording to the CEBM criteria (seeTab. 3 for a description of each gradeof recommendation). If the majorityof articles showed a statistically sig-nificant difference between groups,the results were considered consis-tent across studies for that outcomemeasure. Likewise, if half of the arti-cles that reported on an outcomemeasure showed a significant differ-ence and the other half did not reach

significance, the results were consid-ered inconsistent.

ResultsAbstracts of initially identified arti-cles (n�1,501) were screened foreligibility. The Figure displays oursearch strategy, and Table 1 lists theresults of the Ovid/MEDLINE elec-tronic search. There was one arti-cle22 that, from the title, seemed tomeet our inclusion criteria; how-ever, we were unable to obtain theabstract or full text to determine eli-gibility for inclusion, and no contactinformation was available for theauthors. Accordingly, we were ableto obtain 8 studies in full text thatmet our inclusion criteria. Two ofthe 8 included studies—Holdsworthet al13 and Webster et al14—investi-gated different outcomes from thesame population and cohorts, so this

Table 3.Data Synthesis

Outcome Measure SummaryCEBM Grade of

Recommendationa

Cost One level 3 study and 2 level 4 studies showed significantlydecreased cost in the direct access group vs thephysician referral group; 1 study (level 3) did not reportsignificance, but reported means show a large effect size

B

No. of physical therapy visits 3 level 4 studies and 1 level 3 study showed significantlydecreased visits in the direct access group vs thephysician referral group; 2 studies (levels 2 and 3)showed no significant differences between groups

C

Pharmacological interventions 3 studies (2 level 3 studies, 1 level 4 study) showedsignificantly more use of pharmacological interventions inthe physician referral group vs the direct access group

B

Imaging ordered All 3 studies (2 level 3 studies, 1 level 4 study) showedsignificantly increased imaging ordered in the physicianreferral group vs the direct access group

B

General practitioner, consultationservices, or hospital admits

2 studies (1 level 3 study, 1 level 4 study) showedsignificantly fewer GP visits after physical therapydischarge and significantly fewer hospital admissionsduring physical therapy care; 2 studies (both level 3)showed no difference between groups

D

Patient satisfaction 2 studies (level 3) reported significantly greater satisfactionin the direct access group vs the physician referral group

B

Discharge outcomes (function/goals) and harm

3 studies (2 level 3 studies, 1 level 4 study) show improveddischarge outcomes for direct access vs physician referral;1 study (level 3) showed no difference

1 study showed no reports of harm in either group

C

a Description of grades of recommendation are according to the Centre for Evidence-Based Medicine criteria: A, consistent level 1 studies; B, consistent level2 or 3 studies or extrapolations from level 1 studies; C, level 4 studies or extrapolations from level 2 or 3 studies; D, level 5 evidence or troublinglyinconsistent or inconclusive studies of any level.





review summarized the results from7 datasets reported across the 8included studies. The most commonreasons for exclusion were lack ofanalysis on our outcomes of interestand no direct comparison of directaccess and referral interventions, asspecified in the “Method” section ofthe article.

Table 2 lists characteristics of eachstudy included in this review and thelevel of evidence using the CEBMcriteria (levels ranged from 3 to 4).None of the studies which met ourcriteria were randomized; 4 werenonrandomized prospective cohortstudies, and 4 were nonrandomizedretrospective cohort studies. TheDowns and Black checklist scoresare reported in Table 4 and rangedfrom 13 to 22 out of a total of 26points. Interrater reliability for theDowns and Black checklist scoring(H.A.O. and R.S.S.) was kappa�.931(P�.001; Cohen kappa�.025 stan-dard error). Any differences in ratingwere resolved through consensus.Results were summarized qualita-tively by outcome measures(included below) and are presentedin further detail in Table 2. Datasynthesis results are presented inTable 3.

Health Care CostsData from the included studies sup-ported a grade B recommendationthat costs to patient or insurancecompanies per physical therapy epi-sode of care were less when patientssaw a physical therapist directlyversus through physician referral,likely due to less imaging ordered,injections performed, and medica-tions prescribed. Four studies9,11,13,15

reported on cost differencesbetween direct access and physicianreferral groups, and all reportedlower costs (to the patient, insur-ance company, or health system) inthe direct access group during theparticipants’ episode of care. Pend-ergast et al11 found the mean allow-able amounts during the episode ofphysical therapy care were approxi-mately $152 less for physical thera-py–related costs and $102 less fornon–physical therapy–related costs,amounting to over $250 less for totalcosts per episode of care (P�.001).Mitchell and de Lissovoy9 found thatpaid claims per episode of care were$1,232 less in the direct accessgroup for all services and drugs perepisode of physical therapy care(P�.001). Hackett et al15 reported amean difference of approximately

£38 (�$59) less cost* per patientamong those who incurred costsfrom physical therapy (P�.01; 95%confidence interval�£12.41, £63.65);however, this finding was largelybecause the referral practice had ahigh percentage of patients whoreceived private physical therapytreatments (description of privatephysical therapy not fully explainedin the article). Finally, althoughHoldsworth et al13 did not run statis-tical analyses, patients in the directaccess group had approximately £22(�$34) less cost (not including costto patients), which we extrapolatedwould amount to an average costbenefit to the National Health Ser-vice of Scotland of approximately £2million per year (�$3,107,400).

Number of Physical TherapyVisitsData were available to support agrade C recommendation that thenumber of physical therapy visitswas significantly less in the directaccess group compared with thephysician-referred group. Six articlescompared mean number of physical

* For all conversions, we used Great Britainsterling pound to US dollar, Bank of Englanddaily rate as of August 15, 2013 (http://www.bankofengland.co.uk/boeapps/iadb/Rates.asp).

Table 4.Methodological Quality

Study 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27D&B

Scorea

Hackett et al,15 1993 Y N Y Y P Y N N N N Y Y Y U U Y Y N Y N Y N N Y N 13/26

Holdsworth et al,13 2007 Y Y Y Y N Y Y N N Y Y Y Y N N Y Y Y Y Y Y N N Y Y 18/26

Holdsworth andWebster,12 2004

Y Y Y Y Y Y Y N N Y Y N Y N N Y Y N Y Y Y N N N N 16/26

Webster et al,14 2008 Y Y Y Y P Y Y N N Y N N Y N Y Y Y Y N Y Y N N N Y 16/26

Leemrijse et al,8 2008 Y Y Y Y Y Y Y N Y Y N Y Y Y Y Y Y Y Y Y Y N N Y Y 22/26

Mitchell andde Lissovoy,9 1997

Y Y Y Y N Y U N Y N Y Y Y Y N Y Y Y Y Y Y N N Y Y 18/26

Pendergast et al,11 2012 Y Y Y Y Y Y Y N Y Y Y Y Y Y U Y Y Y Y Y Y N N Y Y 22/26

Moore et al,10 2005 Y Y N Y N Y Y Y N N U U Y Y N Y Y U Y Y Y N N N N 13/26

a Criteria are based on Downs and Black checklist (Appendix 1): Y (yes)�criterion met, N (no)�criterion not met, P�criterion partially met, and U�criterionunable to determine from the study manuscript. Criterion 5 has a maximum of 2 points, and all other criteria have a maximum of 1 point. U ratingsreceived zero points. Criteria 17 and 27 were omitted due to reasons explained in the “Quality Assessment” section.





therapy visits per patient episodeof care with 4 studies (levels 3 and4)8,9,11,12 reporting that patients inthe direct access group had signifi-cantly fewer visits and 2 studies(level 3)13,15 reporting no significantdifference between groups. Of note,both studies conducted in the UnitedStates9,11 that collected data on num-ber of visits showed a significantdifference between groups. Pender-gast et al,11 who included the largestnumber of participants (direct accessgroup, N�17,362; physician referralgroup, N�44,755) of the 6 studies,reported a mean difference of 1.1visits between groups (P�.001).Mitchell and de Lissovoy9 reportedthe largest mean difference, withthe direct access group using 20.2visits compared with the physicianreferral group using 33.6 (P�.0001);however, this study was conductedin 1997, so it might not reflect morerecent practice patterns.

Adjunctive Testing andInterventions (Imaging andPharmacological Interventions)There was a grade B recommenda-tion that less adjunctive testing andfewer interventions were prescribedwhen a patient received physicaltherapy through direct access com-pared with physician referral. All 3studies9,13,15 looking at pharmacolog-ical interventions showed significantdifferences between groups. Mitch-ell and de Lissovoy9 reported therewere significantly fewer drug claimsin the direct access group (P�.01),Hackett et al15 reported fewer med-ications were prescribed in thedirect access group (P�.001), andHoldsworth et al13 reported 12% lesstook nonsteroidal anti-inflammatorydrugs or analgesics in the directaccess group (P�.0001).

All 3 studies 9,13,15 investigating imag-ing showed significant differencesbetween groups. Mitchell and deLissovoy9 reported fewer radiologyclaims (P�.01), Hackett et al15

reported 8% less imaging orderedin the direct access group (no statis-tical tests run), and Holdsworth etal13 showed 6% less imaging orderedin the direct access group (P�.001).

Physician Appointments,Consultation Services, andHospital AdmitsData from the included studies sup-ported a grade D (inconsistent) rec-ommendation that patients in thedirect access group saw their generalpractitioner (GP) or other consul-tants less than in the physician refer-ral group, suggesting that patientsmaintain contact with other medicalproviders despite seeking directaccess to physical therapy. Out of 3studies12,14,15 reporting on frequencyof GP consultation services, onlyHoldsworth and Webster12 found asignificant difference (P�.0113),with 29% of the direct access grouphaving at least one contact with theirGP for the same diagnosis 3 monthsafter physical therapy versus 46%in the physician referral group (forother mean differences, see Tab. 2).Webster et al14 found that the num-ber of GP consultations 1 monthafter physical therapy was approxi-mately the same in both groups (notsignificant, P�.219). Hackett et al15

investigated the frequency of GPvisits during the course of physicaltherapy care and found patients, onaverage, saw their GP for 2 visitsin both groups. Regarding hospitaladmissions, only the study by Mitch-ell and de Lissovoy9 investigated thisoutcome measure and showed signif-icantly fewer mean hospital admis-sions in the direct access group(P�.01).

Patient SatisfactionThe data of the included studies indi-cated a grade B recommendationthat patients reported a higher levelof satisfaction when they receivedphysical therapy through directaccess versus physician referral. The2 studies14,15 that investigated satis-

faction showed that patients in thedirect access group reported greatersatisfaction compared with patientsin the physician referral group.Webster et al14 showed 5% more ofthe participants in the direct accessgroup were satisfied or very satisfied(P�.001). Hackett et al15 showed9% more of the participants in thedirect access group evaluated man-agement of their condition as aver-age or above average, although itwas difficult to conclude whetherthe level of significance (P�.01)would have been the same if onlydirect access and physician referralgroups were compared because thestudy ran these tests among 3 groups(including one group for which datawas not extracted).

Patient Outcomes and HarmA grade C recommendation was sug-gested by data from the includedstudies that patients receiving phys-ical therapy through direct accessversus referral had better outcomesat discharge. Four studies8,12,13,15

reported on discharge outcomes,and although all of the studiesshowed improved outcomes in thedirect access group, the differencesreached significance in 2 studies8,12

(one level 3, one level 4). Leemrijseet al8 reported that the percentage ofpatients who fully achieved goals atdischarge was 9% more in the directaccess group compared with thephysician referral group (P�.001).Holdsworth and Webster12 reportedthe percentage of patients who fin-ished their course of care was 79% inthe direct access group comparedwith 60% in the physician referralgroup (P�.004), and the percentageof those who achieved their goalswas 15% more in the direct accessgroup compared with a controlgroup (P�.079). In this study, signif-icantly less average pain wasreported at discharge (the directaccess group decreased 3 points onthe visual analog scale and the physi-cian referral group decreased 2.5





points on the visual analog scale)(P�.011), although we questionwhether this is a clinically meaning-ful finding. Hackett et al15 reportedthe mean number of days of workmissed due to the condition was 17days less in the direct access groupcompared with the physician referralgroup, although statistical analyseswere not reported for this differ-ence. Holdsworth and col-leagues12,13 reported no significantdifference between groups regard-ing percentage who achieved goals(the direct access group had, onaverage, 2% more achieved goalscompared with the physician refer-ral group, P�.82).

Data from the included studies indi-cated a grade C recommendationthat individuals seen by a physicaltherapist in a direct access capacitydid not result in harm because onlyone level 4 study reported on thisoutcome measure. Moore and col-leagues10 retrospectively comparedharm between direct access and phy-sician referral groups. There wereno reported adverse events in eithergroup resulting from physical thera-pist diagnosis and management, nocredentials or state licenses modifiedor revoked for disciplinary action,and no litigation cases filed againstthe US government in either groupover a 40-month observation period.The sample sizes in that study werequite large, with 50,799 patientsincluded in the direct access groupand 61,854 patients included in thephysician referral group. It is com-monly thought that physical thera-pists seeing patients in a directaccess capacity would result in over-looking serious diagnoses that couldmimic musculoskeletal presentations,thereby putting the patient’s healthat risk. Contrary to this conception,Moore et al cited samples of diagno-ses identified by physical therapistsin the study, which included Ewingsarcoma, Charcot-Marie tooth dis-ease, fractures, nerve injuries (long

thoracic, suprascapular, and spinalnerve root injuries), posterior lateralcorner sprain, osteochondritis dessi-cans, ankylosing spondylitis, tarsalcoalition, compartment syndrome,and scapholunate instability. Of note,compared with the other studies inthis review that involved civilianphysical therapists, the large major-ity of physical therapists in this studywere military physical therapists,with 8% civilian physical therapists,many with specialized training.Approximately 38% of the physicaltherapists had board-certified train-ing in one or more specialties (eg,orthopedics, neurology, sports), and88% had attended a specialty trainingcourse.

Results SummaryDue to limitations inherent in studydesign, differences in number ofparticipants between groups, andother potentially confounding vari-ables, we believe our most relevantfindings are that patient and healthcare costs were not greater in thedirect access group compared withthe physician referral group. Further-more, these results do not indicatethat patients seen through directaccess received more visits orachieved inferior outcomes com-pared with those who were referredby physicians. Harm was notreported in the majority of the stud-ies; however, one large-scale studyexamining military physical thera-pists showed no harm when individ-uals received direct access physicaltherapy. Given that patients in thedirect access group received fewermedications and less imaging whileachieving similar or superior dis-charge outcomes, the results fromthis review suggest a relativedecreased risk of harm in the directaccess group, potentially due tofewer side effects of medication orless exposure to imaging radiation.

DiscussionThe purpose of this review was todetermine whether health carecosts were less and outcomes wereimproved if individuals receivedphysical therapy care through directaccess compared with physicianreferral. All studies (level 3–4 evi-dence) reporting on cost showeddecreased cost in the direct accessgroup (grade B recommendation),likely due to decreased imaging,number of physical therapy visits,and medications prescribed. Simi-larly, all studies (level 3–4 evidence)showed the same or better dischargeoutcomes (grade C), achieved infewer physical therapy visits (gradeC), with increased satisfaction (gradeB) in the direct access group andwithout any evidence of increasedrisk of harm to the patient (gradeC). This preliminary support forimproved outcomes in the directaccess group potentially could bedue to earlier initiation of physicaltherapy. These observations are con-sistent with prior individual studiesthat collectively support improvedoutcomes for patients and decreasedcosts associated with earlier initia-tion of physical therapy clinical man-agement.23–26 Between-cohort differ-ences in each study were generallysmall in magnitude; however, theycould result in meaningful optimiza-tion of patient outcomes anddecreases in costs when distributedover the large US health care envi-ronment. In summary, findings fromthis systematic review support thesafety, efficacy, and cost-effectivenessof physical therapist services byway of direct access compared withphysician-referred episodes of care.

Little previous work has beenconducted to critically evaluate andsynthesize the literature related tophysical therapy clinical manage-ment obtained through direct access.The previous systematic review onthis topic by Robert and Stevens pub-lished in 19974 examined a related





question, reporting results fromstudies largely conducted within theNational Health Service of the UnitedKingdom. Similar to the results ofthis review, Robert and Stevensfound improved waiting time, recov-ery time, convenience, and costsamong patients receiving physicaltherapy through direct or openaccess. However, in this report, theterms “direct access” and “openaccess” seem to have been definedas expeditious physical therapyreferrals from generalist physicians,such as “on-demand” physical ther-apy clinics, which reflects a gate-keeping model, with the GP initiat-ing the physical therapy referral. Webelieve our review was able to moredirectly focus on results of directaccess physical therapy defined bythe consumer self-referring forphysical therapy. Another reviewrecently published by Desmeulesand colleagues27 focused on physicaltherapists in advanced practice orextended scope roles comparedwith usual care by physicians andother medical providers for patientswith musculoskeletal disorders. Sim-ilar to our findings, the review foundadvanced practice care may be as(or more) beneficial than usual careby physicians in terms of treatmenteffectiveness, use of health careresources, economic costs andpatient satisfaction.

In contrast, in our review, we inves-tigated a group of physical thera-pists, the majority of whom werenot practicing in advanced practiceroles (7 out of 8 studies exclusivelyfocused on physical therapists with-out any special training reportedwho largely held master’s or bache-lor’s degrees), and still found advan-tages in terms of treatment effective-ness, use of resources, economiccosts, and patient satisfaction overinitial physician care. As the annualpercentage of civilians who areboard-certified physical therapy spe-cialists has been steadily increasing,

perhaps the profession is progress-ing in skill level to more frequentlyserve in these advance practiceroles. However, in the interim, ourreview suggests the current basictraining and competency require-ments are sufficient for physical ther-apists without this specialized train-ing to function in a direct accesscapacity.

Various methodological limitationswere identified in the literature,which should be addressed in futurestudies. Dependent variable mea-surements and data reporting wereso heterogeneous that data couldnot be pooled through meta-analyticprocedures. Fewer than half of theincluded studies (3 out of 8) wereconducted in the United States,which is relevant because of theunique payer arrangements and prac-tice regulations involving physicaltherapists in the US health care mar-ket. Furthermore, health care costsvary substantially across countries,thus cost savings and expenditurecannot be generalized. All includedstudies involved an outpatient ortho-pedic practice environment, soother practice areas were under-represented. There was no random-ization of study participants togroups, and blinding of participantswas not conducted. This approachis relevant because, in addition topotentially limiting inferences thatcan be made regarding cause andeffect based on the evidence, thereis a possibility for the influence ofuncontrolled selection bias amongindividuals who self-refer for physi-cal therapy through direct access.

From the results of the logistic regres-sion conducted by Holdsworth andWebster,12 individuals in a singlehealth district (n�679) in Scotlandwho engaged in direct access utiliza-tion of physical therapist serviceswere significantly more likely toreport male sex, younger age, shortersymptom duration, and engagement in

paid employment than individualswho received physical therapy inphysician-referred episodes of care.Subsequently, Leemrijse and col-leagues8 reported as the results of alogistic regression analysis that individ-uals in the Netherlands (n�10,519)who are younger, with higher educa-tional attainment, nonspecific spinesymptoms, recurrent symptoms, andprior treatment by a physical therapistwere significantly more likely to havedirect access to physical therapist ser-vices than individuals who werereferred by a physician. Although thisinformation reflects characteristicsthat may be over-represented in thedirect access group, these findingsalso provide valuable information thatcan be used to guide preparation forphysical therapists to function in adirect access environment. Finally,publication bias and selective report-ing within each study could haveintroduced risk of bias to our summaryof evidence.

The consistent results identifiedacross the several moderate-qualitystudies included in this systematicreview may form a solid basis forpolicy and payment decisions thatwould facilitate delivery of physicaltherapist services through directaccess.8,9,11–15,28 Some form of directaccess to physical therapist servicesis currently available by statute in 47out of 50 states (United States),29 aswell as internationally.8,15 However,self-referral accounts have been esti-mated to account for as little as 6%to 10% of referral volume30 in somedirect access states. One reason forthis limitation is that most third-partypayers do not compensate physicaltherapists for evaluation and man-agement of patients who self-referfor physical therapy.† In addition,direct access is unrecognized as a

† A recommended process for third-partyhealth insurance organizations to calculate theeconomic benefit of consumer direct accessto physical therapy is presented in Appendix2.





covered route of access to physicaltherapy in the United States at thefederal level. These legislators andpayers should consider the potentialfor improved patient outcomes andsignificant health care cost savingsby facilitating more widespreaddirect access to physical therapistservices. A common argument madeby proponents of physician referralagainst more widespread directaccess to physical therapist serviceshas been potential adverse effectson patient safety. However, no sci-entific literature is currently avail-able to support this claim. In theUnited States, the Commission onAccreditation in Physical TherapyEducation (CAPTE) criteria supportthe ability of all physical therapists toengage in the delivery of physicaltherapy through direct access. Fur-thermore, direct access to physicaltherapy is commonplace in manyother countries even though thelarge majority of physical therapistspractice with a bachelor’s or mas-ter’s level education. In the UnitedStates, the large majority of physicaltherapist programs are doctor ofphysical therapy programs; how-ever, a comparatively low percent-age of physical therapists practice ina direct access capacity due to thesevarious barriers.

In conclusion, this review suggeststhat physical therapists practicingin a direct access capacity havethe potential to decrease costs andimprove outcomes in patients withmusculoskeletal complaints withoutprescribing medications and order-ing adjunctive testing that couldintroduce harm to the patient. At aminimum, the results presented inthis report show no evidence ofgreater costs or increased number ofvisits or harm when patients self-refer directly to a physical therapist.Finally, despite self-referring forphysical therapy, it appears thatpatients continue to be engagedwith physicians throughout their

course of care; thus, it is unlikelythat widespread implementation ofdirect access to physical therapywill reduce demand for seeking carefrom other practitioners. The poten-tial benefit of direct access to physi-cal therapy in other practice settingsshould be further explored, as wellas alternate pathways for providinghealth services that take advantageof the safety, efficacy, and cost-effectiveness of direct access physi-cal therapy.

Dr Ojha and Dr Davenport provided con-cept/idea/research design, writing, datacollection, project management, and fundprocurement. All authors provided dataanalysis and consultation (including reviewof manuscript before submission). Theauthors thank Eugene Komaroff and Eliza-beth Frank for reviewing the manuscript.

A platform presentation of this research wasgiven at the Combined Sections Meetingof the American Physical Therapy Associa-tion; February 21–24, 2013; San Diego,California.

This study was funded by the Health ServicesResearch Pipeline established through theAmerican Physical Therapy Association tocover basic supplies and conference feesrelated to the research.

DOI: 10.2522/ptj.20130096

References1 Yelin E, Herrndorf A, Trupin L, Sonneborn

D. A national study of medical care expen-ditures for musculoskeletal conditions: theimpact of health insurance and managedcare. Arthritis Rheum. 2001;44:1160–1169.

2 Adams PF, Hendershot GE, Marano MA.Current estimates from the NationalHealth Interview Survey, 1996. VitalHealth Stat 10. 1999;10:1–203.

3 Agency for Healthcare Research and Qual-ity. MEPS HC-110F codebook 2007 outpa-tient department visits and MEPS HC-110Gcodebook 2007 office-based medical pro-vider visits. Available at: http://meps.ahrq.gov/data_stats/download_data/pufs/h110f/h110fdoc.shtml and http://meps.ahrq.gov/mepsweb/data_stats/download_data/pufs/h110g/h110gdoc.shtml. Accessed February25–26, 2010.

4 Robert G, Stevens A. Should general prac-titioners refer patients directly to physicaltherapists? Br J Gen Pract. 1997;47:314–318.

5 Kelly EW, Bradway LF. A team approach tothe treatment of musculoskeletal injuriessuffered by navy recruits: a method todecrease attrition and improve quality ofcare. Mil Med. 1997;162:354–359.

6 Linton SJ, Hellsing AL, Andersson D. Acontrolled study of the effects of an earlyintervention on acute musculoskeletalpain problems. Pain. 1993;54:353–359.

7 Snow BL, Shamus E, Hill C. Physical ther-apy as primary health care: public percep-tions. J Allied Health. 2001;30:35–38.

8 Leemrijse CJ, Swinkels IC, Veenhof C.Direct access to physical therapy in theNetherlands: results from the first year incommunity-based physical therapy. PhysTher. 2008;88:936–946.

9 Mitchell JM, de Lissovoy G. A comparisonof resource use and cost in direct accessversus physician referral episodes of phys-ical therapy. Phys Ther. 1997;77:10–18.

10 Moore JH, McMillian DJ, Rosenthal MD,Weishaar MD. Risk determination forpatients with direct access to physicaltherapy in military health care facilities.J Orthop Sports Phys Ther. 2005;35:674–678.

11 Pendergast J, Kliethermes SA, FreburgerJK, Duffy PA. A comparison of health careuse for physician-referred and self-referredepisodes of outpatient physical therapy.Health Serv Res. 2012;47:633–654.

12 Holdsworth L, Webster V. Direct accessto physiotherapy in primary care: Now?—and into the future? Physiotherapy. 2004;90:64–72.

13 Holdsworth L, Webster V, McFadyen A.What are the costs to NHS Scotland ofself-referral to physiotherapy? Results of anational trial. Physiotherapy. 2007;93:3–11.

14 Webster V, Holdsworth L, McFadyen A,Little H. Self-referral, access and physio-therapy: patients’ knowledge and atti-tudes—results of a national trial. Physio-therapy. 2008;94:141–149.

15 Hackett G, Bundred P, Hutton J, et al. Man-agement of joint and soft tissue injuriesin three general practices: value of on-sitephysiotherapy. Br J Gen Pract. 1993;43:61–64.

16 Oxford Centre for Evidence-Based Medi-cine Levels of Evidence Working Group.The Oxford 2011 levels of evidence. Avail-able at: http://www.cebm.net/index.aspx?o�5653.

17 Downs SH, Black N. The feasibility of cre-ating a checklist for the assessment ofthe methodological quality both of ran-domised and non-randomised studies ofhealth care interventions. J EpidemiolCommunity Health. 1998;52:377–384.

18 Chudyk AM, Jutai JW, Petrella RJ, Speech-ley M. Systematic review of hip fracturerehabilitation practices in the elderly. ArchPhys Med Rehabil. 2009;90:246–262.

19 Hooper P, Jutai JW, Strong G, Russell-Minda E. Age-related macular degenera-tion and low-vision rehabilitation: a sys-tematic review. Can J Ophthalmol. 2008;43:180–187.



http://meps.ahrq.gov/data_stats/download_data/pufs/h110f/h110fdoc.shtml



http://meps.ahrq.gov/mepsweb/data_stats/download_data/pufs/h110g/h110gdoc.shtml



http://www.cebm.net/index.aspx?o=5653

http://www.cebm.net/index.aspx?o=5653



20 Samoocha D, Bruinvels DJ, Elbers NA,et al. Effectiveness of web-based interven-tions on patient empowerment: a system-atic review and meta-analysis. J Med Inter-net Res. 2010;12:e23.

21 Hoenig JM, Heisey DM. The abuse of pow-er: the pervasive fallacy of power calcula-tions for data analysis. The American Stat-istician. 2001;55(11):1–6.

22 Balachandran SRK, Yang MX, Tan C. Eval-uation of a direct access and fast trackroute to physiotherapy at primary health-care centers in Singapore. PhysiotherapySingapore. 2010;13:32–38.

23 Zigenfus GC, Yin J, Giang GM, FogartyWT. Effectiveness of early physical ther-apy in the treatment of acute low backmusculoskeletal disorders. J Occup Envi-ron Med. 2000;42:35–39.

24 Wand BM, Bird C, McAuley JH, et al. Earlyintervention for the management of acutelow back pain: a single-blind randomizedcontrolled trial of biopsychosocial educa-tion, manual therapy, and exercise. Spine(Phila Pa 1976). 2004;29:2350–2356.

25 Fritz JM, Childs JD, Wainner RS, Flynn TW.Primary care referral of patients with lowback pain to physical therapy: impacton future healthcare utilization and costs.Spine (Phila Pa 1976). 2012;37:2114–2121.

26 Nordeman L, Nilsson B, Moller M, Gun-narsson R. Early access to physical therapytreatment for subacute low back pain inprimary health care: a prospective ran-domized clinical trial. Clin J Pain. 2006;22:505–511.

27 Desmeules F, Roy JS, Macdermid JC, et al.Advanced practice physiotherapy inpatients with musculoskeletal disorders: asystematic review. BMC MusculoskeletDisord. 2012;13:107.

28 Moore JH, Goss DL, Baxter RE, et al. Clin-ical diagnostic accuracy and magnetic res-onance imaging of patients referred byphysical therapists, orthopaedic surgeons,and nonorthopaedic providers. J OrthopSports Phys Ther. 2005;35:67–71.

29 FAQ: Direct access at the state level. Avail-able at: http://www.apta.org/stateissues/directaccess/faqs/. Updated May 11, 2012.Accessed January 5, 2013.

30 McCallum CA, DiAngelis T. Direct access:factors that affect physical therapist prac-tice in the state of Ohio. Phys Ther. 2012;92:688–706.



http://www.apta.org/stateissues/directaccess/faqs/

http://www.apta.org/stateissues/directaccess/faqs/



Appendix 1.Modified Downs and Black Criteria and Scoring Guidelinesa

Number Criteria Scoring Criteria

Reporting

1 Is the hypothesis/aim/objective of the study clearlydescribed?

A point was awarded if the hypothesis aim or objective of the studywas implicitly or explicitly indicated anywhere in the article.

2 Are the main outcomes to be measured clearlydescribed in the introduction or “Method”section?

A point was not awarded if the main outcome to be measured wasfirst mentioned in the “Results” section.

3 Are the characteristics of the patients included inthe study clearly described?

A point was awarded if inclusion or exclusion criteria, or both, wereindicated.

4 Are the interventions of interest clearly described? A point was awarded only when the intervention was clear andspecific. For the purposes of this review, we interpreted “clearand specific” to mean direct mention of groups being directaccess compared with referral with or without further descriptorsof what this constituted.

5 Are the distributions of principal confounders foreach group of participants to be comparedclearly described?

Two points were awarded if a study reported any possibleconfounders (eg, sex ratios, age, comorbidities, severity of injury)that might account for differences between groups clearly intable format. One point was awarded if the study indicated thatgroups were matched for any such demographical variables or ifpotential confounders were mentioned in the text of the articlebut not clearly listed in table format. No points were awarded ifthe study did not report any confounders.

6 Are the main findings of the study clearlydescribed?

A point was awarded if quantitative data were reported for all ofthe main outcome measures indicated in the introduction or“Method” section.

7 Does the study provide estimates of the randomvariability in the data for the main outcomes?

A point was awarded if the interquartile range (for non-normallydistributed data), standard error, standard deviation, orconfidence intervals (for normally distributed data) werereported. If the distribution of the data was not described, weassumed that the estimates used were appropriate, and weanswered “yes” (1 point).

8 Have all of the important adverse events that maybe a consequence of the intervention beenreported?

A point was awarded if any adverse events, unwanted side effects,or lack thereof were explicitly indicated from either referral ordirect access interventions. Although adverse events wereoutcome measures extracted from the studies in this review, webelieved that they also were indicative of comprehensivereporting. A point was not awarded if a study made no mentionof the presence or absence of adverse events (eg, loss of licenseof a therapist, minor or serious side effects of intervention) in thedirect access or physician referral groups.

9 Have the characteristics of patients lost to follow-up been described?

The authors of this tool indicated that this question should beanswered “yes” where there were no losses to follow-up orwhere losses to follow-up were so small that findings would havebeen unaffected by their inclusion. For the purposes of thisreview, a point was awarded if a study explicitly reported thatthere were no losses to follow-up or if the losses to follow-upaccounted for a maximum of 10% of the sample of participantsoriginally enrolled in the study (or up to 5% of the originalnumber of participants assigned to each direct access andphysician referral group). No point was awarded if the study didnot report the number of patients lost to follow-up or thisinformation could not be obtained from the tables, figures, ortext of the study.

10 Have actual probability values been reported (eg,.035 rather than �.05) for the main outcomes,except where the probability value is less than.001?

A point was awarded if the exact P value was provided for bothstatistically significant and nonsignificant results for at least themain outcome measures. A point was not awarded if a studysimply indicated that the results for the main outcome measureswere not significant without providing the exact P value.

(Continued)





Appendix 1.Continued


External Validity

11 Were the individuals asked to participate in thestudy representative of the entire populationfrom which they were recruited?

A point was awarded if the study identified the source populationfor patients and described how the patients were selected.Patients were determined to be representative if they comprisedthe entire source population, an unselected sample ofconsecutive patients, or a random sample (only feasible where alist of all members of the relevant population exists). Where astudy did not report the proportion of the source populationfrom which the patients are derived, the question was answeredas unable to determine.

12 Were those individuals who were prepared toparticipate representative of the entirepopulation from which they were recruited?

The proportion of those asked who agreed to participate orresponded should be stated. Validation that the sample wasrepresentative would include demonstrating that the distributionof the main confounding factors was the same in the studysample and the source population (eg, if the demographics anddiagnoses of the patients were considered representative of atypical outpatient musculoskeletal practice, the study wasawarded a point). No point was awarded if the proportion ofthose asked who agreed to participate or responded was notstated.

13 Were the staff, places, and faculties where thepatients were treated representative of thetreatment the majority of patients receive?

A point was awarded unless the study specifically stated thatpatients were treated by a therapist who received specializedtraining or the direct access or physician referral group receivedtreatment in a specialized facility defined by advanced training orfocus of intervention in “niche” areas of physical therapy (eg,pediatrics).

Internal Validity–Bias

14 Was an attempt made to blind study participantsto the intervention they received?

A point was awarded if the patients were not aware of, or wouldhave no way of knowing (as in the case of retrospective studies),which intervention they received. The study was not awarded apoint if it was prospective and failed to mention whether thepatients had knowledge of whether they were assigned to thedirect access or physician referral group.

15 Was an attempt made to blind those measuringthe main outcomes of the intervention?

A point was awarded if the study specifically stated that thoseassessing the outcome measures were unaware of (or would haveno way of knowing) whether the patients were in the directaccess or physician referral group.

16 If any of the results of the study were based on“data dredging,” was this made clear?

A point was awarded if no retrospective unplanned (at the outset ofthe study) subgroup analyses were reported.

17 In trials and cohort studies, do the analyses adjustfor different lengths of follow-up of patients, or,in case-control studies, is the time periodbetween the intervention and outcome thesame for cases and controls?

For the purposes of this review, this question was omitted due toreasons previously stated.

18 Were the statistical tests used to assess the mainoutcomes appropriate?

If the distribution of the data (normal or not) was not described, itwas assumed that the estimates used were appropriate, and apoint was awarded. No point was awarded for studies thatreported qualitative or quantitative data without any form ofstatistical comparisons or if the statistical tests reported were notappropriate.

19 Was adherence to the intervention reliable? If the authors in prospective studies reported nonadherence tophysical therapy intervention or adherence could not bedetermined, the study was not awarded a point. In retrospectivestudies, data were collected only for those patients whocompleted their episode of care (adherence to physical therapyassumed), and a point was awarded. For studies where the effectof any nonadherence was likely to bias any association to thenull, the study received a point.

(Continued)





Appendix 1.Continued


20 Were the main outcome measures used accurate(valid and reliable)?

A point was awarded if the primary outcome measures werethought to be valid and reliable (eg, number of physical therapyvisits per chart report), regardless of whether reliability or validitywas reported. A point was not awarded if at least one of theprimary outcome measures in the study was not valid or reliableor if this information was not reported or could not bedetermined (ie, a questionnaire without reported validity orreliability).

Internal Validity–Confounding (Selection Bias)

21 Were the patients in different intervention groups(trials and cohort studies), or were the cases andcontrols (case-control studies) recruited from thesame population?

A point was awarded when participants from both direct accessand physician referral groups were recruited from the samepopulation. Otherwise, a point was not awarded (eg, a point wasnot awarded when all participants from the physician referralgroup received care at clinic A and all participants in the directaccess group received care at clinic B, because they could haverepresented 2 distinct populations).

22 Were study participants in different interventiongroups (trials and cohort studies), or were thecases and controls (case-control studies)recruited over the same period of time?

A point was awarded when the study provided a specific time linefor patient recruitment (prospective studies) or when data werecollected between reported dates of patient care (retrospectivestudies).

23 Were study participants randomized tointervention groups?

A point for random allocation was awarded if random allocation ofpatients was stated in the “Method” section of the article. Theprecise method of randomization need not be specified. Quasi-randomization allocation procedures, such as allocation by bedavailability, did not satisfy this criterion. For crossover studydesigns, a point was awarded when participants were randomlyallocated in the order in which treatments were received.

24 Was the randomized intervention assignmentconcealed from both patients and health careteam until recruitment was complete andirrevocable?

The study did not receive a point unless the participants wererandomly allocated and the methods for ensuring randomallocation were specified.

25 Was there adequate adjustment for confoundingin the analyses from which the main findingswere drawn?

A point was awarded unless the effect of the main confounders wasnot investigated or confounding was demonstrated, but noadjustment was made in the final analyses.

26 Were losses of patients to follow-up taken intoaccount?

A point was awarded as long as the number of dropouts lost tofollow-up accounted for less than 10% of the initial number oftotal participants or a maximum of 5% from each group. Thequestion was answered with “unable to determine” if thenumber of patients lost to follow-up were not reported or couldnot be deduced from the outcome data (ie, initial and finalsample sizes not indicated).

Power

27 Did the study have sufficient power to detect aclinically important effect where the probabilityvalue for a difference being due to chance is lessthan .05?

For the purposes of this evidence-based review, this question wasomitted.

a For original criteria, refer to Downs and Black.17





Appendix 2.One Method of Calculating Differences in Cost Between Direct Access and Physician-Referred Episodes of Care

In the event that third-party payers want to calculate differences in cost between direct access and referred episodesof care in their own records, one potential way cited previously11 is listed below:

1. Run analyses on only those members who were continuously enrolled for at least 6 months before until 2 monthsafter treatment.

2. Identify physical therapist services/procedures through billing codes for services (ie, CPT codes 97001–97999and revenue code Y42xx) and provider specialty type for physical therapist.

3. Aggregate physical therapy claims for each member by defining the start of the episode as the date of thephysical therapy initial evaluation code (ie, CPT 97001).

4. Identify the ending date as the last physical therapy claim before a 60-day window with no further physicaltherapy claims (any second initial evaluation within that episode was considered a re-evaluation rather than thestart of a new episode).

5. Verify that all physical therapy visits occurred in a physical therapy office or in a hospital-based outpatient facilitysetting.

6. If an individual had multiple physical therapy episodes of care in the identified time frame, randomly select anepisode for inclusion in the analysis.

7. Classify physical therapy episodes of care as physician-referred or self-referred by generating a list of physician(MD or DO) specialty types who might reasonably refer patients for physical therapy. (If there is any doubt,include specialties so that physician referrals are not misclassified as self-referrals.)

8. Classify as physician-referred if one or more claims from any physician provider on the list occurred within 30days prior to the initial physical therapist evaluation. Otherwise, classify the episode as self-referred.

9. When defining whether the physical therapy episode of care could reasonably be considered initiated by aphysician or not in the 30-day window prior to the first physical therapist evaluation visit, disregard diagnosesreported on the physician claims because a patient might see a physician for one problem and request a physicaltherapy referral at that time for an unrelated medical issue. Consider diagnoses on the physician claims whenlooking for visits with a primary diagnosis that agreed with the diagnosis used by the physical therapist.

10. Consider each date of service a visit for counting total number of visits per episode of care.

11. If claims are date-spanned, as may occur when analyzing outpatient hospital physical therapy claims, determinereasonable number of CPT codes or units of service per visit to calculate the number of visits in the date-spannedperiod.

12. Health care use can be measured by the number of physical therapy visits per episode of care and the totalallowable amounts per visit and for the episode extracted from the claims data. The allowable amount, alsoknown as the allowable fee, the maximum allowable fee, or the usual, customary, and reasonable fee (Blue CrossBlue Shield 2011) is a proxy for health care cost and is defined as the total amount of physical therapy benefitfor the physical therapist services and procedures billed regardless of member responsibility or the contractualrelationship between the payer and the provider of physical therapist services.

13. In the analysis, account for any medical or payment policy that influences referral patterns by physicians orability of patients to self-refer for physical therapy.





doi: 10.2522/ptj.20130096Originally published online September 12, 2013

2014; 94:14-30.PHYS THER. Heidi A. Ojha, Rachel S. Snyder and Todd E. DavenportTherapy Episodes of Care: A Systematic ReviewDirect Access Compared With Referred Physical

References

http://ptjournal.apta.org/content/94/1/14#BIBLfor free at: This article cites 27 articles, 6 of which you can access

Cited by

http://ptjournal.apta.org/content/94/1/14#otherarticles

This article has been cited by 1 HighWire-hosted articles:

Information Subscription http://ptjournal.apta.org/subscriptions/

Permissions and Reprints http://ptjournal.apta.org/site/misc/terms.xhtml

Information for Authors http://ptjournal.apta.org/site/misc/ifora.xhtml

at APTA Member on March 2, 2014http://ptjournal.apta.org/Downloaded from

http://ptjournal.apta.org/content/94/1/14#BIBL

http://ptjournal.apta.org/content/94/1/14#BIBL



http://ptjournal.apta.org/subscriptions/

http://ptjournal.apta.org/site/misc/terms.xhtml

http://ptjournal.apta.org/site/misc/ifora.xhtml



originally published online september 12, 2013 - … directaccess ptj 1-14.pdforiginally published...

Documents