CENTRAL, PEDro, PubMed, and

EMBASE Are the Most Comprehensive

Databases Indexing Randomized
Controlled Trials of Physical Therapy
Interventions
Zoe A. Michaleff, Leonardo O.P. Costa, Anne M. Moseley, Christopher G. Maher,
Mark R. Elkins, Robert D. Herbert, Catherine Sherrington
Background. Many bibliographic databases index research studies evaluating the
effects of health care interventions. One study has concluded that the Physiotherapy
Evidence Database (PEDro) has the most complete indexing of reports of randomized
controlled trials of physical therapy interventions, but the design of that study may
have exaggerated estimates of the completeness of indexing by PEDro.
Objective. The purpose of this study was to compare the completeness of index-
ing of reports of randomized controlled trials of physical therapy interventions by 8
bibliographic databases.
Design. This study was an audit of bibliographic databases.
Methods. Prespecied criteria were used to identify 400 reports of randomized
controlled trials from the reference lists of systematic reviews published in 2008 that
evaluated physical therapy interventions. Eight databases (AMED, CENTRAL,
CINAHL, EMBASE, Hooked on Evidence, PEDro, PsycINFO, and PubMed) were
searched for each trial report. The proportion of the 400 trial reports indexed by each
database was calculated.
Results. The proportions of the 400 trial reports indexed by the databases were as
follows: CENTRAL, 95%; PEDro, 92%; PubMed, 89%; EMBASE, 88%; CINAHL, 53%;
AMED, 50%; Hooked on Evidence, 45%; and PsycINFO, 6%. Almost all of the trial
reports (99%) were found in at least 1 database, and 88% were indexed by 4 or more
databases. Four trial reports were uniquely indexed by a single database only (2 in
CENTRAL and 1 each in PEDro and PubMed).
Limitations. The results are only applicable to searching for English-language
published reports of randomized controlled trials evaluating physical therapy
interventions.
Conclusions. The 4 most comprehensive databases of trial reports evaluating
physical therapy interventions were CENTRAL, PEDro, PubMed, and EMBASE. Clini-
cians seeking quick answers to clinical questions could search any of these databases
knowing that all are reasonably comprehensive. PEDro, unlike the other 3 most
complete databases, is specic to physical therapy, so studies not relevant to physical
therapy are less likely to be retrieved. Researchers could use CENTRAL, PEDro,
PubMed, and EMBASE in combination to conduct exhaustive searches for randomized
trials in physical therapy.
Z.A. Michaleff, PT, is a PhD candi-
date, The George Institute for
Global Health, The University of
Sydney, PO Box M201, Missenden
Rd, Sydney, New South Wales
2050, Australia. Address all corre-
spondence to Ms Michaleff at:
zmichaleff@georgeinstitute.org.au.
L.O.P. Costa, PT, PhD, is Associate
Professor, Universidade Cidade de
Sao Paulo, Sao Paulo, Brazil.
A.M. Moseley, PhD, is Senior
Research Fellow, The George Insti-
tute for Global Health, The Univer-
sity of Sydney.
C.G. Maher, PT, PhD, is Director,
Musculoskeletal Division, The
George Institute for Global Health,
and Professor, Sydney Medical
School, The University of Sydney.
M.R. Elkins, PT, PhD, is Research
Physiotherapist, Department of
Respiratory Medicine, Royal Prince
Alfred Hospital, Camperdown,
New South Wales, Australia.
R.D. Herbert, PhD, is Senior
Research Fellow, Musculoskeletal
Division, The George Institute for
Global Health, and Associate Pro-
fessor, Sydney Medical School,
The University of Sydney.
C. Sherrington, PhD, is Senior
Research Fellow, The George Insti-
tute for Global Health, The Univer-
sity of Sydney.
[Michaleff ZA, Costa LOP, Moseley
AM, et al. CENTRAL, PEDro,
PubMed, and EMBASE are the most
comprehensive databases indexing
randomized controlled trials of
physical therapy interventions. Phys
Ther. 2011;91:190197.]
2011 American Physical Therapy
Association
Research Report
Post a Rapid Response to
this article at:
ptjournal.apta.org
190 f Physical Therapy Volume 91 Number 2 February 2011
E
vidence-based practice involves
the integration of the best
research evidence with clinical
expertise and patient values to make
optimal health care decisions for
individual patients.
1
An evidence-
based approach requires clinicians
and researchers to be able to access
the best evidence available quickly
and easily.
2
Advances in electronic
technologies (eg, the Internet) and
electronic bibliographic databases
have facilitated access to information
and made searching, locating, and
retrieving evidence easier and less
time-consuming for health care pro-
fessionals, including physical thera-
pists.
3
However, both practitioners
asking clinical questions and
researchers conducting exhaustive
searches for systematic reviews
experience several challenges when
searching databases. Challenges
encountered by clinicians can
include limited access because of
high subscription costs, rudimentary
searching skills (so that searching
may retrieve a large proportion of
irrelevant literature and not identify
key evidence), and time con-
straints.
35
These issues can result in
full text on the net bias as clini-
cians preferentially retrieve articles
that are easily accessed.
6
Research-
ers, however, need to conduct com-
prehensive and unbiased searches to
identify all studies relevant to a spe-
cic topic.
6,7
Conducting such
searches often requires both knowl-
edge of and access to multiple
databases.
8
Many electronic databases index
research about the effects of health
care interventions. The databases
that are most commonly used in sys-
tematic reviews of physical therapy
interventions are the Medical Litera-
ture Analysis and Retrieval System
Online (MEDLINE), Excerpta Medica
Database (EMBASE), Cochrane
Library (including Cochrane Review
Group Registers and Cochrane Cen-
tral Register of Controlled Trials
[CENTRAL]), Clinical Index of Nurs-
ing and Allied Health Literature
(CINAHL), PsycINFO, Allied and
Complementary Medicine Database
(AMED), and Physiotherapy Evi-
dence Database (PEDro).
9
The data-
bases differ with respect to primary
professional target audience (eg,
physicians, nurses, psychologists,
physical therapists), journals
indexed, and types of articles
indexed (eg, all research articles or
just randomized controlled trials and
systematic reviews).
4
Variations in
indexing mean that a clinician may
have more success in retrieving rel-
evant literature by searching a spe-
cic database that comprehensively
indexes particular types of research
from the clinicians eld. Alterna-
tively, a researcher may be able to
perform exhaustive searches with
fewer but more comprehensive
databases.
Moseley et al
4
recently compared
8 bibliographic databases (AMED,
CENTRAL, CINAHL, EMBASE, Hooked
on Evidence, PEDro, PsycINFO, and
PubMed) for completeness of index-
ing of reports of randomized con-
trolled trials of physical therapy inter-
ventions and concluded that PEDro
(99%) and CENTRAL (98%) provided
the most complete coverage. The
authors acknowledged that their study
may have obtained optimistic esti-
mates of the completeness of PEDro
and CENTRAL because both databases
systematically extract reports of ran-
domized controlled trials that are
included in Cochrane systematic
reviews
10,11
and because the pool of
trials used to compare the 8 databases
also was obtained from the reference
lists of Cochrane reviews. In the
present study, we sought to address
this potential limitation by comparing
the databases using an independent
pool of trials generated from system-
atic reviews evaluating physical ther-
apy interventions published in 2008
and indexed by PubMed. This strategy
reduces the risk of selection bias
because the link between the 2 data-
bases (PEDro and CENTRAL) and
PubMed is not as direct as the link
with Cochrane reviews.
Specically, our objective was to
compare the completeness of index-
ing of reports of randomized con-
trolled trials of physical therapy
interventions by 8 bibliographic
databases (AMED, CENTRAL, CINAHL,
EMBASE, Hooked on Evidence, PEDro,
PsycINFO, and PubMed) using an
independent pool of trials.
Method
Dening the Pool of Trials
A PubMed search was conducted to
identify all systematic reviews that
evaluated physical therapy interven-
tions and were published in 2008.
The PubMed Clinical Queries nd
systematic reviews function was
applied to the search strategy:
(Physical Therapy Modalities
[Mesh] OR Physical Therapy (Spe-
cialty)[Mesh]) AND systematic[sb]
AND(2008/01/01[PDAT]:2008/12/
31[PDAT]). Systematic reviews were
considered to be relevant if they eval-
uated the effects of a physical therapy
intervention
12
and were published in
2008. The title and, if necessary, the
abstract and full text were screened
independently by 2 authors with a
background in physical therapy
(Z.A.M. and L.O.P.C.) to identify rele-
vant reviews. The same 2 authors also
coded each relevant review for area of
Available With
This Article at
ptjournal.apta.org
eAppendix: Systematic Reviews
Used to Generate the Pool of 400
Reports of Randomized
Controlled Trials
Audio Abstracts Podcast
This article was published ahead of
print on December 9, 2010, at
ptjournal.apta.org.
Comprehensive Databases for Physical Therapy Trials
February 2011 Volume 91 Number 2 Physical Therapy f 191
physical therapy (cardiothoracics,
continence and womens health, ergo-
nomics and occupational health, ger-
ontology, musculoskeletal, neurology,
orthopedics, pediatrics, and sports).
Any disagreements were resolved by
discussion.
The eligible systematic reviews were
sorted in random order by use of the
random number function in
Microsoft Ofce Excel.* The refer-
ence lists of these systematic reviews
were extracted until a pool of 400
unique reports of randomized con-
trolled trials was obtained. This pro-
cess ensured that a random sample
of systematic reviews would be used
to generate the pool of trial reports,
regardless of how many reviews
were required to generate a pool of
400 trial reports.
The reports of randomized con-
trolled trials were eligible for inclu-
sion if they met the following 4 cri-
teria: the article was a full report
published in a peer-reviewed jour-
nal, at least 2 interventions were
compared (1 of these interventions
could be a no-treatment control or a
sham treatment), at least 1 of the
interventions being evaluated was
part of physical therapist practice,
12
and the study involved random allo-
cation or intended-to-be-random allo-
cation of participants to interven-
tions. These criteria were applied to
focus the analysis on full reports of
randomized controlled trials evaluat-
ing physical therapy interventions.
Searching the Bibliographic
Databases
Eight bibliographic databases were
searched to determine if they indexed
each of the 400 unique reports of ran-
domized controlled trials. The data-
bases searched were AMED via OVID
(http://www.ovid.com via The Uni-
versity of Sydney subscription), CEN-
TRAL via the Wiley InterScience portal
(http://www3.interscience.wiley.com),
CINAHL via EBSCOhost (http://
www.ebscohost.comvia The University
of Sydney subscription), EMBASE via the
subscription version (http://www.
embase.com/), Hooked on Evidence
(http://www.hookedonevidence.org),
PEDro (http://www.pedro.org.au),
PsycINFO via OVID, and PubMed
(http://www.ncbi.nlm.nih.gov/
pubmed). Seven of these databases
were selected because they have been
shown to be the most commonly used
for systematic reviews of physical ther-
apy interventions.
9
HookedonEvidence
also was searched because it is an evi-
dence resource developed specically
for physical therapists. These databases
represent both subscription-only and
open-access databases and both generic
and physical therapyspecic databases.
Two authors (Z.A.M. and A.M.M.) per-
formed all searches using combinations
of author names, words in the title, jour-
nal names, volume numbers, and page
numbers. If a report could not be
located in a database, the search was
repeated independently by another
author (L.O.P.C.). All searches were per-
formed between December 14, 2009,
and December 20, 2009. The total num-
ber of entries each database was
recorded on the day on which the
search was completed.
Analyses
For each database, the number of
trial reports that could be identied
from the pool of 400 reports was
recorded. Condence intervals for
the proportions were computed
with the Wilson score method.
13
The
number of trial reports indexed
uniquely by each database and the
number of trial reports indexed by 2
or more databases also were deter-
mined. This step involved identifying
the combination of databases index-
ing each trial report and determining
the number of times each combina-
tion occurred.
Role of the Funding Source
The study team designed, con-
ducted, and reported this trial inde-
pendent of the sources of funding.
* Microsoft Corp, One Microsoft Way, Red-
mond, WA 98052-6399.
356 systematic reviews evaluating
physical therapy interventions
identified in the PubMed search
132 relevant systematic reviews
580 references extracted from the
first 39 systematic reviews
400 full-text randomized controlled
trials evaluating physical therapy
interventions included in analysis
224 ineligible systematic reviews
Reasons for exclusion:
- Not published in 2008 (n=25)
- Did not evaluate physical therapy
intervention (n=199)
180 ineligible trials:
- Random allocation not used (n=72)
- Physical therapy intervention not
evaluated (n=50)
- Study population not treated in
physical therapy practice (n=23)
- Duplicates (n=22)
- Referencing error/not full published
article (n=13)
Figure.
Selection of the pool of trials.
Comprehensive Databases for Physical Therapy Trials
192 f Physical Therapy Volume 91 Number 2 February 2011
Results
Pool of Trials
The PubMed search retrieved 356
systematic reviews (Figure). Of
these, 132 systematic reviews were
relevant because they evaluated the
effects of a physical therapy inter-
vention and were published in 2008.
The pool of 400 unique reports of
randomized controlled trials was
compiled fromthe rst 39 systematic
reviews from the randomly sorted
list of reviews (listed in the eAppen-
dix; available at ptjournal.apta.org).
These reviews evaluated physical
therapy interventions in 8 of the 9
areas of physical therapy (there were
no reviews for ergonomics and
occupational health) (Tab. 1). The
majority of the reviews evaluated
physical therapy interventions for
musculoskeletal conditions (10 of
the 39 reviews). Seven of the
reviews were published in the
Cochrane Database of Systematic
Reviews. All but 2 of the reviews
were published in English.
The pool of 400 unique reports of
randomized controlled trials was
derived from the 580 references
included in the 39 systematic
reviews. A total of 180 references
were excluded because they did not
meet the eligibility criteria for
reports of randomized controlled tri-
als. Specically, 72 of these refer-
ences were excluded because they
were reports of other research meth-
ods (eg, case control study) or did
not contain between-group analyses
(eg, protocols and secondary analy-
ses) (Figure). The 400 eligible trial
reports were published between
1971 and 2008, and the majority
were published in English (384
reports). The list of trial reports
included in the analysis is available
from the corresponding author.
Completeness of Indexing by the
Databases
Four databases indexed large propor-
tions of the pool of reports of ran-
domized controlled trials: CENTRAL
(95%), PEDro (92%), PubMed (89%),
and EMBASE (88%). A further 3 data-
bases indexed approximately half of
the pool of trial reports: CINAHL
(53%), AMED (50%), and Hooked on
Evidence (45%). PsycINFO indexed a
small proportion (6%). Table 2
shows the quantity of trial reports
indexed by each of the bibliographic
databases, as well as the total num-
ber of entries for each database.
PubMed (19,379,156 articles
indexed) and EMBASE (13,402,910
articles) were the largest databases
(Tab. 2). Of the physical therapy
specic databases, PEDro (15,678
articles) was larger than Hooked on
Evidence (5,517 articles).
Almost all of the reports of random-
ized controlled trials in the pool of
trials (394/400, 99%) were found in
at least 1 of the databases surveyed,
and 88% of the trial reports were
indexed by 4 or more databases
(Tab. 3). The 6 trial reports that were
not indexed by any database were all
published in Chinese-language jour-
nals: Chinese Journal of Athletic
Medicine, Traditional Chinese Med-
Table 1.
Number of Systematic Reviews Included
in This Study by Area of Physical
Therapy
Area of Physical Therapy
No. of
Systematic
Reviews
Musculoskeletal 10
Cardiothoracics 5
Gerontology and
musculoskeletal
5
Neurology 4
Pediatrics 4
Cardiothoracics and
gerontology
2
Sports and orthopedics 2
Continence and womens
health
2
Sports 2
Gerontology 1
Gerontology and neurology 1
Orthopedics 1
Ergonomics and occupational
health
0
Total 39
Table 2.
Reports of Randomized Controlled Trials From a Pool of 400 Trials Indexed by
Various Databases and Total Number of Entries for Each Database
a
Database
No. (%) of
Trials
Indexed
95% Condence
Interval of
Percentage
Total Number
of Entries
CENTRAL 378 (95) 9296 608,405
PEDro 366 (92) 8894 15,678
PubMed 356 (89) 8692 19,379,156
EMBASE 352 (88) 8491 13,402,910
CINAHL 210 (53) 4857 2,168,147
AMED 200 (50) 4555 238,165
Hooked on Evidence 181 (45) 4050 5,517
PsycINFO 22 (6) 48 2,914,911
a
The data are ranked according to the completeness of indexing. CENTRALCochrane Central
Register of Controlled Trials, PEDroPhysiotherapy Evidence Database, EMBASEExcerpta Medica
Database, CINAHLClinical Index of Nursing and Allied Health Literature, AMEDAllied and
Complementary Medicine Database.
Comprehensive Databases for Physical Therapy Trials
February 2011 Volume 91 Number 2 Physical Therapy f 193
icine Research, Chinese Journal of
Modern Developments in Tradi-
tional Medicine (Zhong Xi Yi Jie He
Za Zhi), Chinese Journal of
Rehabilitation (Zhong Guo Kang
Fu Yi Xue Za Zhi), Journal of Bei-
jing Sport University (Beijing Ti Yu
Da Xue Xue Bao), and Chinese
Journal of Sports Medicine (Zhong-
guo Yun Dong Yi Xue Za Zhi). Four
trial reports were uniquely indexed
by 1 database only: 2 in CENTRAL, 1
in PEDro, and 1 in PubMed (Tab. 3).
Four trial reports were uniquely
indexed by a single database only (2
in CENTRAL and 1 each in PEDro
and PubMed).
Larger proportions of trial reports
were retrieved when combinations
of the 4 most complete databases
(CENTRAL, PEDro, PubMed, and
EMBASE) were searched. Searching
any pair of these 4 databases
retrieved 96% or more of the pool of
reports of randomized controlled
trials. The combination of CENTRAL
and PEDro indexed the largest pro-
portion of the pool of trial reports
(391/400), whereas the smallest pro-
portion of trial reports was found
when the combination of EMBASE
and PubMed was used (383/400).
The most efcient and exhaustive
search (ie, the minimum number of
databases required to be searched to
retrieve the maximum number of
trial reports) involved searching
CENTRAL and PEDro with the addi-
tion of either EMBASE or PubMed,
which retrieved 393 of 400 or 394 of
400 trial reports, respectively.
Discussion
In the present study, we evaluated
the completeness of indexing of
reports of randomized controlled tri-
als of physical therapy interventions
by 8 bibliographic databases. CEN-
TRAL, PEDro, PubMed, and EMBASE
were identied as the most compre-
hensive databases indexing this inde-
pendent pool of randomized con-
trolled trials. CENTRAL, PEDro, and
PubMed all indexed trials that
appeared in that single database
only. Searching CENTRAL and PEDro
in combination with either EMBASE
or PubMed retrieved the largest num-
ber of trial reports.
The results of the present study were
similar to those of an earlier study by
Moseley et al
4
; they also identied
CENTRAL, PEDro, PubMed, and
EMBASE as the 4 most comprehen-
sive databases indexing reports of tri-
als of physical therapy interventions.
However, to attempt to reduce the
risk of selection bias, we used a dif-
ferent method to generate the pool
of reports of randomized controlled
trials for evaluating the complete-
ness of indexing by the various data-
bases. Moseley et al
4
used the trials
included in 30 Cochrane reviews of
physical therapy interventions, a
strategy that may have favored CEN-
TRAL and PEDro because both of
those databases systematically
extract trials directly from Cochrane
reviews.
10,11
In contrast, we included trials from
the reference lists of a random sam-
ple of systematic reviews of physical
therapy interventions published in
2008 and indexed by PubMed, a
strategy that decreased the link
between the pool of trials and the
databases under investigation.
Although our strategy reduced the
risk of selection bias, we acknowl-
edge that several other sources of
bias may affect the indexing of trials
by bibliographic databases. These
include language bias (trials with sta-
tistically signicant results are more
likely to be published in English-
language journals),
14
publication
bias (research with statistically signif-
icant results is more likely to be sub-
mitted and published than research
with nonsignicant ndings),
15,16
and database bias (journals pub-
lished in English in developed coun-
tries are more likely to be indexed by
electronic databases than those in
developing countries; for example,
of the 46 physical therapy journals
listed on the World Confederation
of Physical Therapy Web site,
17,18
MEDLINE indexes 15 of the 31
English-language journals but none
of the 15 nonEnglish-language jour-
nals). The inuences of these poten-
tial sources of bias are difcult to
quantify.
Because clinicians and researchers
have different needs, they may
choose to use different databases.
For clinicians, a major consideration
may be the accessibility of the data-
base (eg, open access or subscrip-
tion only). Of the 4 most complete
databases (CENTRAL, PEDro, PubMed,
and EMBASE), 3 (CENTRAL,
19
PubMed,
20
and PEDro
21
) are freely
available. In the present study, we
evaluated the advanced search mode
of EMBASE, which is available only by
subscription. The simple search mode
of EMBASE is freely available but was
not evaluated in the present study.
PEDro is unique (compared with the
other 3 most comprehensive data-
bases) because it is a physical therapy
specic database that indexes only
randomized controlled trials, system-
atic reviews, and clinical practice
guidelines. Therefore, PEDro may be a
good rst choice for clinicians search-
ing for reports of randomized con-
trolled trials of physical therapy
interventions.
Researchers have a greater need to
conduct comprehensive and unbi-
ased searches. Consequently, it often
is necessary for researchers to search
multiple databases. The 39 system-
atic reviews used in the present
study searched an average of 6 data-
bases (range319) to identify the
studies included in their analyses. In
the present study, we concluded
that researchers can conduct an
exhaustive literature search for trials
evaluating physical therapy interven-
tions using as few as 3 databases
(CENTRAL and PEDro in combina-
Comprehensive Databases for Physical Therapy Trials
194 f Physical Therapy Volume 91 Number 2 February 2011
Table 3.
Bibliographic Databases Indexing Each Report of a Randomized Controlled Trial From the Pool of 400 Trials
a
No. of Databases
Indexing the Trial
No. (%)
of Trials
Indexed
No. of Trials for
Each Combination
of Databases AMED CENTRAL CINAHL EMBASE
Hooked
on
Evidence PEDro PsycINFO PubMed
0 6 (1.5) 0
1 4 (1.0) 2
1
1
2 5 (1.3) 3
2
3 32 (8.0) 9
6
5
3
2
2
2
2
1
4 78 (19.5) 51
6
4
3
3
2
2
2
1
1
1
1
1
5 97 (24.2) 28
25
19
8
3
3
2
2
2
1
1
1
1
1
(Continued)
Comprehensive Databases for Physical Therapy Trials
February 2011 Volume 91 Number 2 Physical Therapy f 195
tion with either EMBASE or
PubMed). Six reports could not be
located by the 8 databases used in
the present study. All were
published in Chinese-language jour-
nals and evaluated the effects of tai
chi. These may have been found by
searching, in addition, a Chinese-
languagespecic database such as
China Academic Journal Full-Text
Database (http://china.eastview.com/
kns50/Navigator.aspx?ID1). In gen-
eral, it may be necessary to search par-
ticular databases, in addition to CEN-
TRAL, PEDro, PubMed, and EMBASE,
if an intervention is specic to a cer-
tain culture or is outside mainstream
clinical practice.
The limitations of the present study
include the degree to which the
results can be generalized to
searches for randomized controlled
trials published in languages other
than English. Six trial reports could
not be found despite the use of var-
ious combinations of author names,
article titles, journal names, volumes,
issues, and years of publication. This
result may indicate that these trial
reports were not indexed by these
databases, or it may have been due to
indexing errors. Furthermore, the 8
databases scrutinized in the present
study were not evaluated for their
ability to identify gray literature (ie,
conference proceedings, published
abstracts, theses, and books), which
may be important sources of trial
reports for systematic reviewers.
Finally, these databases were not
evaluated for the sensitivity (ie, the
proportion of relevant references in
the database that are retrieved
22(p19)
)
or the specicity (ie, the proportion
of retrieved references that are rele-
vant to the search
22(p19)
) of reports
retrieved with key word searches
alone.
In conclusion, CENTRAL, PEDro,
PubMed, and EMBASE are the most
complete bibliographic databases for
reports of randomized controlled tri-
als evaluating physical therapy inter-
ventions. Researchers could use
these databases in combination to
conduct exhaustive searches for
reports of trials of physical therapy
interventions. We recommend a
search of CENTRAL and PEDro in
combination with either EMBASE or
PubMed. Clinicians seeking rapid
answers to questions about the
effects of physical therapy interven-
tions could use any 1 of the 4 most
comprehensive databases (3 of
which are freely available).
Dr Costa, Dr Moseley, Dr Maher, Dr Elkins,
and Dr Sherrington provided concept/idea/
research design. Ms Michaleff, Dr Costa, Dr
Moseley, Dr Maher, Dr Herbert, and Dr Sher-
rington provided writing. Ms Michaleff, Dr
Costa, and Dr Moseley provided data collec-
Table 3.
Continued
No. of Databases
Indexing the Trial
No. (%)
of Trials
Indexed
No. of Trials for
Each Combination
of Databases AMED CENTRAL CINAHL EMBASE
Hooked
on
Evidence PEDro PsycINFO PubMed
6 91 (22.7) 29
28
23
2
2
2
1
1
1
1
1
7 84 (21.0) 77
5
1
1
8 3 (0.8) 3
Total 400 (100)
a
AMEDAllied and Complementary Medicine Database, CENTRALCochrane Central Register of Controlled Trials, CINAHLClinical Index of Nursing and
Allied Health Literature, EMBASEExcerpta Medica Database, PEDroPhysiotherapy Evidence Database.
Comprehensive Databases for Physical Therapy Trials
196 f Physical Therapy Volume 91 Number 2 February 2011
tion. Ms Michaleff, Dr Costa, Dr Moseley,
and Dr Herbert provided data analysis. Dr
Maher provided project management. Dr
Moseley, Dr Maher, and Dr Sherrington pro-
vided fund procurement. Dr Maher provided
facilities/equipment. Ms Michaleff, Dr Costa,
Dr Moseley, Dr Maher, Dr Elkins, and Dr
Sherrington provided consultation (includ-
ing reviewof manuscript before submission).
Some of the authors are developers of the
Physiotherapy Evidence Database (PEDro).
PEDro is funded by the Motor Accidents
Authority of New South Wales; the Motor
Accidents Insurance Commission (Queens-
land, Victoria, Australia); the Victorian Trans-
port Accidents Commission (Victoria, Austra-
lia); and physical therapy associations in
Australia, Austria, Canada, Denmark, Fin-
land, Germany, Ireland, the Netherlands,
New Zealand, Portugal, Singapore, Spain,
Sweden, Switzerland, and the United States.
Ms Michaleff holds a postgraduate award
from the Australian government. Professor
Maher, Associate Professor Herbert, and Dr
Sherrington hold research fellowships
funded by the National Health and Medical
Research Council of Australia.
This article was submitted April 1, 2010, and
was accepted October 11, 2010.
DOI: 10.2522/ptj.20100116
References
1 Straus S, Richardson W, Glasziou P,
Haynes R. Evidence-Based Medicine: How
to Practice and Teach EBM. 3rd ed. Edin-
burgh, Scotland: Elsevier Churchill Living-
stone; 2005.
2 McGowan JL, Grad R, Pluye P, et al. Elec-
tronic retrieval of health information by
healthcare providers to improve practice
and patient care. Cochrane Database Syst
Rev. 2009;3:CD004749.
3 Herbert R, Jamtvedt G, Mead J, Hagen K.
Practical Evidence-Based Physiotherapy.
Edinburgh, Scotland: Elsevier; 2005.
4 Moseley AM, Sherrington C, Elkins MR,
et al. Indexing of randomised controlled
trials of physiotherapy interventions: a
comparison of AMED, CENTRAL,
CINAHL, EMBASE, Hooked on Evidence,
PEDro, PsycINFO and PubMed. Physio-
therapy. 2009;95:151156.
5 Maher CG, Sherrington C, Elkins M, et al.
Challenges for evidence-based physical
therapy: accessing and interpreting high-
quality evidence on therapy. Phys Ther.
2004;84:644654.
6 Murali N, Murali H, Auethavekiat P, et al.
Impact of FUTON and NAA bias on visibil-
ity of research. Mayo Clin Proc. 2004;79:
10011006.
7 Crumley ET, Wiebe N, Cramer K, et al.
Which resources should be used to iden-
tify RCT/CCTs for systematic reviews: a
systematic review. BMC Med Res Meth-
odol. 2005;5:24.
8 Betra n AP, Say L, Gulmezoglu AM, et al.
Effectiveness of different databases in
identifying studies for systematic reviews:
experience from the WHO systematic
review of maternal morbidity and mortal-
ity. BMC Med Res Methodol. 2005;5:6.
9 Moseley AM, Elkins MR, Herbert RD, et al.
Cochrane reviews use more rigorous
methods than non-Cochrane reviews: sur-
vey of systematic reviews in physio-
therapy. J Clin Epidemiol. 2009;62:
10211030.
10 Physiotherapy Evidence Database. Fre-
quently asked questions. Available at:
http://www.pedro.org.au/english/faq/
#question_one. Updated October 11,
2010. Accessed February 22, 2010.
11 The Cochrane Library. Cochrane Central
register of controlled trials. Available at:
http://www.thecochranelibrary.com/view/
0/AboutTheCochraneLibrary.html#CENTRAL.
Accessed February 22, 2010.
12 Guide to Physical Therapist Practice. 2nd
ed. Phys Ther. 2001;81:3950, 105129,
139311, 313469, 471593, 595680.
13 Altman D. Statistics With Condence:
Condence Intervals and Statistical
Guidelines. London, United Kingdom:
BMJ Books; 2000.
14 Egger M, Zellweger-Zahner T, Schneider
M, et al. Language bias in randomised con-
trolled trials in English and German. Lan-
cet. 1997;347:326329.
15 Hopewell S, Loudon K, Clarke MJ, et al.
Publication bias in clinical trials due to
statistical signicance or direction of trial
results. Cochrane Database Syst Rev.
2009;1:MR000006.
16 Easterbrook P, Berlin J, Gopalan R, Mat-
thews D. Publication bias in clinical
research. Lancet. 1991;337:867872.
17 World Confederation for Physical Thera-
py. ISPJE membership: journalsdescrip-
tions of physiotherapy journals (Excel for-
mat, searchable). Available at: http://
www. wcpt . org/i spj e/membershi p.
Updated October 15, 2010. Accessed
October 15, 2010.
18 Egger M, Smith G. Bias in location and
selection of studies. BMJ. 1998;316:
6166.
19 The Cochrane Collaboration. Cochrane.org
search. Available at: http://www3.inter
science.wiley.com/cgi-bin/mrwhome/1065
68753/AccessCochraneLibrary.html.
Accessed October 15, 2010.
20 National Center for Biotechnology Infor-
mation (NCBI). PubMed help. Available at:
http://www.ncbi.nlm.nih.gov/bookshelf/
br.fcgi?bookhelppubmed&partpub
medhelp#pubmedhelp.FAQs. Updated
October 15, 2010. Accessed February 22,
2010.
21 Physiotherapy Evidence Database. Home
page. Available at: http://www.pedro.org.
au/. Updated October 11, 2010. Accessed
October 15, 2010.
22 Brazier H, Murphy AW, Lynch C, Bury G.
Searching for the evidence in pre-hospital
care: a review of randomised controlled
trials. J Accid Emerg Med. 1999;16:1823.
Comprehensive Databases for Physical Therapy Trials
February 2011 Volume 91 Number 2 Physical Therapy f 197