[ research report ]

ANTHONY G. SCHNEIDERS, PT, PhD1 • S. JOHN SULLIVAN, PhD2 • PAUL A. HENDRICK, PT, PhD3 • BENJAMIN D.G.M. HONES, PT, BPhty4
ANDREW R. MCMASTER, PT, BPhty4 • BRIDGET A. SUGDEN, PT, BPhty5 • CELIA TOMLINSON, PT, BPhty6

The Ability of Clinical Tests to Diagnose

Stress Fractures: A Systematic
Review and Meta-analysis

S
tress fractures are a bone-related overuse injury commonly seen with different sports and activities; how-
in athletes and military personnel.29,35 This condition was first ever, 80% to 95% of stress fractures oc-
cur in the lower limbs,14,25,40,43,44,46 with the
reported in the literature in 1855 by a Prussian military phys-
tibia being the most commonly injured
ician7 and first described in athletes in 1958.44 Stress fractures bone, accounting for approximately 50%
have been suggested to account for approximately 10% of all athletic of all cases.1
injuries35,41,44 and are most commonly seen in runners, with an incidence Clinically, lower-limb stress fractures
of up to 20% in this population.30 The site of the stress fracture varies can be difficult to diagnose, due to a wide
range of potential differential diagnoses16
that include compartment syndrome,
TTSTUDY DESIGN: Systematic literature review criteria. Meta-analysis was used to statistically
soft tissue injuries, infection, and other
and meta-analysis. analyze the data extracted from the ultrasound
overuse conditions such as medial tibial
TTOBJECTIVES: To evaluate the diagnostic ac- articles and demonstrated a pooled sensitivity of
64% (95% confidence interval [CI]: 55%, 73%), stress syndrome and periostitis.43,44 The
curacy of clinical tests to identify stress fractures
in the lower limb. specificity of 63% (95% CI: 54%, 71%), positive diagnosis of a stress fracture is contingent
on a detailed clinical examination14 that
TTBACKGROUND: Stress fractures are a bone-
likelihood ratio of 2.1 (95% CI: 1.1, 3.5), and
related overuse injury primarily occurring in
negative likelihood ratio of 0.3 (95% CI: 0.1, 0.9). incorporates the patient history35,41,43 and
Tuning fork test data could not be pooled; however, contributing risk factors,5,24,32,33,40 as well
the lower limb and commonly affecting running
athletes and military personnel. Physical examina- sensitivity, specificity, positive likelihood ratio, and as a thorough physical examination,35,41,43
tion procedures and clinical tests are suggested negative likelihood ratio ranged from 35% to 92%, and is most often confirmed with radio-
for diagnosing stress fractures; however, data on 19% to 83%, 0.6 to 3.0, and 0.4 to 1.6, respectively.
logical imaging.5
the diagnostic accuracy of these tests have not TTCONCLUSION: The results of this systematic Radiological imaging for stress
been investigated through a systematic review of review do not support the specific use of ultra-
the literature.
fractures has traditionally included
sound or tuning forks as standalone diagnostic
roentgenograms (plain radiographs),18
TTMETHODS: A systematic review was conducted tests for lower-limb stress fractures. As the overall
scintigraphy (bone scan), magnetic reso-
in 8 electronic databases to identify diagnostic diagnostic accuracy of the tests investigated is not
accuracy studies, published between January 1950 nance imaging (MRI), and computed
strong, based on the calculated likelihood ratios, it
and June 2011, that evaluated clinical tests against is recommended that radiological imaging should tomography. Although plain radiographs
a radiological diagnosis of lower-limb stress continue to be used for the confirmation and are commonly used as initial reference
fracture. Retrieved articles were evaluated using diagnosis of stress fractures of the lower limb. standards, they have limited useful-
the Quality Assessment of Diagnostic Accuracy
Studies tool, and a meta-analysis was performed TTLEVEL OF EVIDENCE: Diagnosis, level 1a–. ness due to their inability to detect bony
J Orthop Sports Phys Ther 2012;42(9):760-771, changes during the early development of
where appropriate.
a stress fracture,3,8,13,15,16,28,29,35,40,41,43 and
TTRESULTS: Nine articles investigating 2 clinical
Epub 19 July 2012. doi:10.2519/jospt.2012.4000
TTKEY WORDS: diagnosis, lower limb, tuning fork,
computed tomography is not frequently
procedures, therapeutic ultrasound (n = 7) and
tuning fork testing (n = 2), met the study inclusion ultrasound, validity used due to low diagnostic sensitivity.8,41
The gold standard for stress fracture

Senior Lecturer, Centre for Physiotherapy Research, School of Physiotherapy, University of Otago, Dunedin, New Zealand. 2Professor, Centre for Physiotherapy Research, School
1

of Physiotherapy, University of Otago, Dunedin, New Zealand. 3Lecturer, Division of Physiotherapy Education, University of Nottingham, Nottingham, UK. 4Physical Therapist,
School of Physiotherapy, University of Otago, Dunedin, New Zealand. 5Physical Therapist, Physio Direct, Adelaide, Australia. 6Physical Therapist, Physiotherapy Department,
Southland Hospital, Invercargill, New Zealand. Address correspondence to Dr A.G. (Tony) Schneiders, School of Physiotherapy, University of Otago, PO Box 56, Dunedin 9054,
New Zealand. E-mail: tony.schneiders@otago.ac.nz t Copyright ©2012 Journal of Orthopaedic & Sports Physical Therapy

760  |  september 2012  |  volume 42  |  number 9  |  journal of orthopaedic & sports physical therapy

42-09 Schneiders.indd 760 8/21/2012 4:14:00 PM

 diagnosis is either triple-phase techne-
tium-99m bone scan (scintigraphy)35,40,41
or MRI.36,51 Scintigraphy is able to diag-
nose stress fractures by identifying areas
of increased uptake of the radioisotope3
and has a reported sensitivity of 100%
and specificity of 76%.8,41 MRI has been
shown to have comparable sensitivity3,41
and superior specificity to scintigraphy
for assessment of bone pathology.41 Al-
though MRI is less invasive and provides
greater anatomic detail than scintigra-
phy,16,35 it is more costly.3,43 Of further
concern is the accessibility of these pro-
cedures to many patients.43
Although radiological imaging can
be used to confirm suspected stress
fractures in a primary care setting,
other common clinical tests have been
suggested to show diagnostic poten-
tial.13,18,26,27,29,31,36,51 The use of therapeu-
tic ultrasound for the diagnosis of stress
fractures has been proposed as a relative-
ly accessible and effective tool for physi-
cal therapists.8,11,35,36,43 Numerous studies
suggest that the elicitation of pain as the
ultrasound is applied over the fracture
site8,11,29,35,43 may be a good diagnostic in-
dicator of an underlying stress fracture.
Tuning forks have also been suggested
as effective diagnostic clinical tools, with
application of the tuning fork to an iden-
tified area of maximal tenderness overly-
ing a suspected stress fracture suggested
to provoke pain due to irritation of the
damaged periosteum.26,41
Systematic reviews of diagnostic accu-
racy studies focus on providing validity
measures of a test and, where possible,
pooling data through a meta-analysis to
offer clinicians a summary of evidence
of the test’s diagnostic accuracy. To de-
termine the diagnostic accuracy and
validity of a clinical test, specific Bayes-
ian probability and performance metrics
should be examined. These include the
number of false positives and negatives
associated with the test, together with
its sensitivity, specificity, positive and
negative likelihood ratios (+LR and –LR,
respectively), and the diagnostic odds
ratio (DOR).34 The sensitivity of a test
journal of orthopaedic & sports physical therapy  |  volume 42  |  number 9  |  september 2012  |  761
42-09 Schneiders.indd 761
is its ability to obtain a positive result
when the condition is actually present;
inversely, specificity is the ability of that
test to obtain a negative result when the
condition is absent. 34 Likelihood ratios
correct the true positive rate by the false
positive rate, revealing how much more
likely it is that a person has the diagnosis
after an index test is completed.34 Likeli-
hood ratios have an advantage over sen-
sitivity, specificity, and predictive values
because they are independent of pathol-
ogy prevalence and therefore applicable
across settings and patients.34 The DOR
is a statistic used to provide a summary of
the performance of diagnostic tests and
augment the comparison between study
results. It combines sensitivity, specificity,
and +LR and –LR to convey how much
greater the probability is that a positive
test result will indicate the event of inter-
est, for example, the presence of a stress
fracture.12
Although anecdotal and selective
research evidence has for decades sug-
gested the ability of nonradiological tests
to diagnose stress fractures, it remains
unclear whether this actually is the case.
The purpose of this study was to system-
atically review the literature and apply
meta-analysis procedures to diagnostic
studies, where appropriate, to establish
which clinical tests have the best accuracy
in musculoskeletal and orthopaedic clini-
cal practice to diagnose stress fractures.
Common clinical testing procedures and
modalities, such as ultrasound and tun-
ing forks, were considered for evaluation
in this review.
METHODS
Search Strategy
A
n initial systematic search of
the literature was conducted on
May 13, 2011, and the search strat-
egy results were monitored until June
13, 2011. A comprehensive search with
no language restrictions was conducted
covering the period 1950 to 2011 in the
following databases: AMED, CINAHL,
Embase, MEDLINE, PEDro, PubMed,
Scopus, and SPORTDiscus. Key words/
terms were identified after a search for
synonyms associated with the term stress
fracture. These terms were entered into
the selected databases in various combi-
nations to identify a range of clinical tests
employed in the examination of suspect-
ed stress fractures, as outlined in FIGURE
1. The Medical Subject Heading (MeSH)
database search included the following
terms: stress fracture*, fatigue fracture*,
insufficiency fracture*, march fracture*,
diagnos*, investigat*, identif *, analys*,
physical exam*, interpret*, palpat*, pres-
sure*, and ultrasound*.
Studies were considered for inclusion
if they reported on the use of clinical (in-
dex) tests to diagnose stress fracture and
were published as full reports before June
13, 2011. The inclusion criteria required
that articles (1) report 1 or more index
tests, (2) utilize at least 1 radiological
reference test, (3) report or allow com-
putation of diagnostic values (sensitivity,
specificity, +LR, and –LR), (4) not impose
an age restriction for participants, (5)
were published between 1950 and 2011,
and (6) include only lower-limb stress
fractures. Articles specifically investigat-
ing pathological stress fractures or stud-
ies not conducted on human subjects
were excluded.
Each database was searched by 2 in-
dependent reviewers (B.A.S. and C.T.) for
titles that met the inclusion criteria. The
2 reviewers then evaluated the retrieved
abstracts of each article for possible in-
clusion based on the criteria specified
above. References from all included ar-
ticles were manually searched for addi-
tional relevant articles. The full text of all
articles included after the manual search
was examined to determine inclusion in
the systematic review. A third reviewer
(A.G.S.) was consulted if an initial con-
sensus could not be reached.
Diagnostic Accuracy Statistics
Articles investigating the diagnostic ac-
curacy of clinical binary classification
tests to diagnose radiologically con-
firmed stress fractures were required to
8/21/2012 4:14:02 PM
 [ research report
report, or allow calculation of, sensitiv-
ity, specificity, +LR, –LR, and the DOR
for each included test.19 We chose not to
report positive and negative predictive
values because they are strongly affected
by the prevalence of the condition, which
we believe has not been adequately es-
tablished for lower-limb stress fractures.
A 2-by-2 contingency table was created
to identify the true positive, false posi-
tive, true negative, and false negative re-
sults, as reported in the original articles.
These specific Bayesian probability and
performance metrics were selected due
to their ability to aid in the interpre-
tation of clinical data extracted from
diagnostic studies. The clinical interpre-
tations of likelihood ratios for this study
are defined in TABLE 1. However, there
is no current accepted taxonomy for
characterizing magnitudes of sensitivity
and specificity. For the purposes of this
study, we have arbitrarily defined these
as low if 50% or less, low to moderate if
between 51% and 64%, moderate if be-
tween 65% and 74%, moderate to high if
between 75% and 84%, and high if 85%
or greater.
Quality Assessment
The methodological quality of the in-
cluded articles was assessed indepen-
dently by 2 reviewers (A.R.M. and
B.D.H.) using the Quality Assessment of
Diagnostic Accuracy Studies (QUADAS)
tool developed by Whiting et al,50 which
is frequently used to evaluate the qual-
ity of diagnostic studies.2 The review-
ers were blinded to the authors, date of
publication, and journals in which the
articles were published. The QUADAS
tool is composed of 14 items individually
scored as either “yes,” “no,” or “unclear.”
Nine items relate to bias, 3 items to the
quality of the reporting, and 2 items to
variability. The reviewers familiarized
themselves with the QUADAS tool and
discussed the quality items of the scor-
ing system prior to the evaluation. This
allowed for more uniform interpretation
of each study and diminished quality as-
sessment bias. The reviewers indepen-
762  |  september 2012  |  volume 42  |  number 9  |  journal of orthopaedic & sports physical therapy
42-09 Schneiders.indd 762
]
Search strategy Databases
1. stress fracture* OR fatigue • MEDLINE
fracture* OR bone stress OR bone • CINAHL
strain* OR insufficiency fracture* • SPORTDiscus
OR repetitive stress injur* OR • PubMed
march fracture* • Embase
2. diagnosis* OR investigat* OR • Scopus
confirm* OR identif* OR clinical • AMED
test* OR analys* OR physical • PEDro
exam* OR interpret* OR diagnostic
tool* OR test*
3. tuning fork* OR vibration fork* OR Electronic database
tap* OR palpat* OR squeeze* OR search, n = 9321
tender* OR focal tender* OR
pressure* OR ultrasound* OR
therapeutic ultrasound* OR heel Included after title
drop OR jarring test OR markle test screen, n = 739
OR drop test OR bump test OR
percussion test OR tap* test*
Duplicates
4. (1 and 2)
removed, n = 210
5. (3 and 4)
Full abstracts, n = 529
Excluded after abstract
screen, n = 514
Hand-searched reference
list, abstracts Full texts, n = 15
reviewed, n = 11
Full texts, n = 4
Full texts, n = 19
Excluded after full-text
screen, n = 10
Included in review, n = 9
Quality assessment
FIGURE 1. Search and selection strategy.
dently scored each of the 14 criterion QUADAS Scoring
items as yes or no when studies either For this systematic review, the QUA-
satisfied or failed to meet the criteria. DAS scoring system proposed by the
Items were scored as unclear when in- original developers50 was utilized to as-
sufficient information was provided to sess the quality and rate the validity of
categorize an item. In the case of dis- the included studies. Using this system,
agreement between reviewers, a third item weightings based on and scaled
reviewer (A.G.S.) adjudicated to assign for potential bias or variation were used
a single score for each study. to develop the scoring for the retrieved
8/21/2012 4:14:03 PM
 is otherwise called the Woolf-Haldane
Clinical Interpretation correction (for the odds ratio). 37 Forest
TABLE 1
of Likelihood Ratios 23 plots and confidence intervals (CIs) were
generated, when appropriate (FIGURES 2
Positive Likelihood Ratio Negative Likelihood Ratio Shift in Probability Condition through 6).
>10 <0.1 Large, often conclusive
5-10 0.1-0.2 Moderate but usually important RESULTS
2-5 0.2-0.5 Small, sometimes important

T
1-2 0.5-1 Very small, rarely important he search strategy process and
result are outlined in FIGURE 1. The
initial electronic database search
Study Name n Confidence Interval
yielded a total of 9321 articles. After ex-
amining titles for context and key words,
Boam (1996) 78 0.429 (0.268, 0.605)
739 articles were subsequently selected
for possible inclusion. After title dupli-
Devereaux (1984) 18 0.533 (0.276, 0.778)
cations were removed, 529 abstracts
Giladi (1984) 106 0.745 (0.601, 0.853)
were screened based on the inclusion
and exclusion criteria, with 15 titles re-
Lowdon (1986) 32 0.900 (0.670, 0.983) tained for further analysis. A hand search
of each of these articles revealed 11 ad-
Moss (1983) 19 0.909 (0.574, 0.996) ditional studies of interest, of which 4
were deemed relevant following abstract
Nitz (1980) 54 1.000 (0.875, 0.997) review. After full-text examination, 9
articles4,13,18,26,27,29,31,36,51 met the inclu-
Romani (2000) 26 0.000 (0.014, 0.487) sion criteria and were assessed using the
QUADAS tool.50 The number of articles
Overall 333 0.642 (0.548, 0.727)
retrieved during this review was limited,
and it is recognized that rating scores
can potentially affect conclusions, based
0.0 0.2 0.4 0.6 0.8 1.0
on the quality of estimates of diagnostic
FIGURE 2. Sensitivity and 95% confidence interval data for therapeutic ultrasound to diagnose stress fractures in accuracy.49 Due to these factors, studies
the lower limb. were not partitioned relative to score or
excluded from the review or subsequent
studies. Items 1, 5, 10, 11, and 12 were were included and assumed to be accu- meta-analysis due to their QUADAS
scored 3 points for yes; items 3 and 6 rate. The data were then independently score.
were scored 2 points for yes; and all other reviewed by a second reviewer (B.D.H.).
items (2, 4, 7-9, 13, and 14) were scored Where appropriate, MetaAnalyst Ver- Study Characteristics and
1 point for yes. Items were scored zero if sion 3.13 beta, developed by Wallace et Methodological Quality
the response was no or unable to be de- al,47 was used to statistically analyze the The 9 articles retained after the system-
termined (unclear). The highest possible data extracted from included studies. atic search investigated either therapeu-
score achievable, if all criteria were met The method in which MetaAnalyst cal- tic ultrasound (n = 7) or tuning fork tests
in all categories, was 26. culates the overall test accuracy indices (n = 2) to diagnose stress fractures of
is well described in the literature.47 As the lower limb. Two different reference
Data Extraction and Quantitative previously indicated, we chose to pre- (criterion) tests (MRI and bone scintig-
Synthesis sent sensitivity, specificity, +LR, –LR, raphy) were used throughout these stud-
Data were extracted by 1 author and DOR in the meta-analysis. Where ies, with 1 study also including the use of
(A.R.M.), who was experienced in me- values of zero are identified in contin- roentgenogram51 as a comparison to the
ta-analyses, for calculation of sensitivity gency tables, MetaAnalyst adds 0.5 to all clinical test. Participants in the studies
and specificity for all clinical diagnostic cells in the contingency table to popu- reviewed ranged from 18 to 78 in num-
tools examined. If raw data were not late the cell, allowing full computation ber and from 19 to 31 years of age. The
provided, the sensitivity and specificity and synthesis of data and calculation age of participants was not specified in
values calculated by the original authors of diagnostic accuracy statistics. This 2 studies.27,31 Male participants only were

journal of orthopaedic & sports physical therapy  |  volume 42  |  number 9  |  september 2012  |  763

42-09 Schneiders.indd 763 8/21/2012 4:14:04 PM

 [ research report ]
detail.
Study Name n Confidence Interval
The total score, item score, and per-
centage agreement between the 2 re-
Boam (1996) 78 0.488 (0.336, 0.644)
viewers who performed the quality
Devereaux (1984) 18 0.000 (0.028, 0.696)
assessment using the QUADAS tool are
presented in TABLE 3. All items had an
Giladi (1984) 106 0.673 (0.532, 0.790) initial reviewer agreement ranging from
55.6% (item 4) to 100% (items 1, 5, 7-10,
Lowdon (1986) 32 0.750 (0.430, 0.934) and 14). Items 2 and 12 had 66.7% agree-
ment, items 3 and 13 had 77.8% agree-
Moss (1983) 19 1.000 (0.601, 0.987) ment, and items 6 and 11 had 88.9%
agreement.
Nitz (1980) 54 0.800 (0.559, 0.935) Sensitivity, specificity, +LRs, and –LRs
and their associated 95% CIs were calcu-
Romani (2000) 26 1.000 (0.801, 0.995)
lated from the study data if they were
not specifically provided in the original
Overall 333 0.631 (0.542, 0.712)
articles,13,27,29,31 and all data were recal-
culated where possible for verification
0.0 0.2 0.4 0.6 0.8 1.0 purposes, even if provided in the original
articles.4,18,26,36,51
FIGURE 3. Specificity and 95% confidence interval data for therapeutic ultrasound to diagnose stress fractures in The QUADAS tool identified criteria
the lower limb. considered important to the method-
ological quality of the retrieved studies.
For instance, the study by Lowdon27 did
Study Name n Confidence Interval
not clearly define the participant charac-
Boam (1996) 78 1.074 (0.627, 1.719)
teristics, which might have given rise to
a spectrum bias, and many of the studies
Devereaux (1984) 18 1.405 (0.548, 2.805) did not clearly specify the selection crite-
ria for included subjects. The reference
Giladi (1984) 106 2.273 (1.567, 3.312) tests used also varied between studies.
Nitz and Scoville31 used plain radiogra-
Lowdon (1986) 32 2.968 (1.621, 5.563) phy, a reference test that is considered
less sensitive and specific than MRI and
Moss (1983) 19 3.670 (1.732, 8.178) scintigraphy. Lesho26 and Wilder et al51
had a significantly long time between in-
Nitz (1980) 54 3.331 (2.138, 5.065)
dex and reference testing (range, 0-432
days), which might have introduced bias
Romani (2000) 26 1.714 (0.333, 4.398)
due to a potential change in the pathology
Overall 333 2.092 (1.099, 3.517)
phase or status. However, in all cases, the
reference standard was independent of
the index tests and was always described
0.1 1.0 10.0 in sufficient detail to permit its replica-
tion. This was not always the case with
FIGURE 4. Positive likelihood ratio and 95% confidence interval data for therapeutic ultrasound to diagnose stress the index test, which was not clearly de-
fractures in the lower limb. scribed in a number of articles.4,18,31 In all
of the studies, except that by Boam et al,4
examined in 2 studies,18,31 male and fe- personnel or recreational/competitive it was unclear whether the index and/or
male participants were examined in 6 athletes, and all were suspected of having reference tests were completed without
studies,4,26,27,29,36,51 and, in the remaining stress fractures of the lower limb, the ma- the examiner’s knowledge of other test
study, Devereaux et al,13 the sex of par- jority of which were located in the tibia. results. In most studies, there were insuf-
ticipants was not specified. The majority TABLE 2 describes the study characteristics ficient or unclearly defined clinical data
of study participants were either military and participant demographics in further reported on each participant, compared

764  |  september 2012  |  volume 42  |  number 9  |  journal of orthopaedic & sports physical therapy

42-09 Schneiders.indd 764 8/21/2012 4:14:05 PM

 to those that would have been available in
Study Name n Confidence Interval
a normal clinical setting. All of the stud-
ies, except those by Moss and Mowat29
Boam (1996) 78 1.043 (0.665, 1.559)
and Wilder et al,51 clearly reported un-
interpretable and/or intermediate test Devereaux (1984) 18 1.336 (0.330, 5.857)
results and withdrawals from the study,
thus negating biased assessment of the Giladi (1984) 106 0.368 (0.208, 0.570)
test characteristics and test performance
in the majority of studies. Lowdon (1986) 32 0.168 (0.036, 0.390)

Pooled Analysis Moss (1983) 19 0.155 (0.021, 0.405)

TABLE 4 depicts the pooled sensitivity, spec-
ificity, +LR, –LR, and DOR values for the Nitz (1980) 54 0.055 (0.002, 0.166)

total sample of ultrasound participants (n

Romani (2000) 26 0.911 (0.545, 1.175)
= 333) included in the meta-analysis, and
FIGURES 2 through 6 depict these values in
Overall 333 0.350 (0.079, 0.861)
forest plots. Due to the 100% sensitiv-
ity of ultrasound testing reported in the
study by Nitz and Scoville,31 an accurate 0.01 0.10 1.00 10.00
DOR could not be calculated and was
therefore not included in tabulated or FIGURE 5. Negative likelihood ratio and 95% confidence interval data for therapeutic ultrasound to diagnose stress
pooled results. All other meta-analytical fractures in the lower limb.
statistics were calculated for all included
studies examining ultrasound. Our analy- Study Name n Confidence Interval
sis of the diagnostic ability of ultrasound
demonstrated low to moderate pooled Boam (1996) 78 1.078 (0.372, 2.457)
sensitivity (64%; 95% CI: 55%, 73%) and
specificity (63%; 95% CI: 54%, 71%). Devereaux (1984) 18 2.025 (0.118, 7.272)
Data pertaining to the tuning fork
test were not pooled using meta-analysis, Giladi (1984) 106 6.646 (2.741, 14.100)
due to an insufficient number of studies
meeting the inclusion criteria for this Lowdon (1986) 32 26.950 (3.609, 114.600)
review (n = 2). Lesho26 conducted the
tuning fork test using a 128-Hz tuning Moss (1983) 19 171.200 (3.794, 539.900)

fork, whereas Wilder et al51 investigated

Romani (2000) 26 3.372 (0.024, 16.420)
the diagnostic ability of forks of 3 dif-
ferent frequencies (128 Hz, 256 Hz, and
Overall 279 6.199 (0.697, 22.750)
512 Hz) against 3 radiological reference
tests. The 256-Hz tuning fork test, com-
pared to radiography, MRI, and bone 1 10 100 1000
scintigraphy, had the highest sensitivity
values (92.3%, 90.0%, and 77.7%, re- FIGURE 6. Diagnostic odds ratio and 95% confidence interval data for therapeutic ultrasound to diagnose stress
spectively), the lowest specificity values fractures in the lower limb.

(19.3%, 20.0%, and 25.0%, respectively),

and +LRs between 1.04 and 1.14. The DISCUSSION sis can lead to increased morbidity and
study by Wilder et al51 scored poorly on possible progression to complete frac-

the QUADAS, with a score of 12/26. The
study by Lesho26 had a QUADAS score of
22/26 and reported 75% sensitivity and
67% specificity, with a +LR of 2.3 for
the 128-Hz tuning fork used to diagnose
tibial stress fractures.
S
tress fractures of the lower
limb are a common injury that can
be difficult to diagnose and trouble-
some to manage. Early detection is con-
sidered vital to provide appropriate and
informed management, as a late diagno-
ture. Moreover, the ability to rule out a
stress fracture is of significant clinical
benefit.
Although diagnosis can often be con-
firmed with radiological means, it is not
always practical or accessible. Referral

journal of orthopaedic & sports physical therapy  |  volume 42  |  number 9  |  september 2012  |  765

42-09 Schneiders.indd 765 8/21/2012 5:18:29 PM

 [ research report ]

TABLE 2 Study Characteristics and Participant Demographics

Authors Index Tests and Parameters Subjects Assessors Reference Standard Definition of the Positive Index Tests
Boam et al4 Ultrasound; 2.0 W/cm2; 30-s 78 patients (87% men) who were 1 sonography technician, Scintigraphy Discomfort or pain with active
application referred for bone scintiscan to rule nuclear medicine de- ultrasound application
out tibial stress fracture (mean age, partment; 1 uninvolved
24 y) party (result tabulating)
Devereaux et al13 Ultrasound; 3-cm head; 0.75 18 patients/athletes attending the Thermogram reader; Scintigraphy Very unpleasant sensation of intense
MHz; intensity increasing sports injury clinic with signs and 1 physiotherapist; pressure or pain with ultrasound
from 0 to 2.0 W/cm2 symptoms of stress fracture of the 1 nuclear medicine application
tibia or fibula (mean age, 23 y; range, physician
18-37 y)
Giladi et al18 Ultrasound; 2.0 W/cm2 53 male soldiers with suspected stress Not specified Roentgenogram; Complaint of acute pain or a
fracture of the tibia (mean age, 19 y; scintigraphy stabbing sensation with
range, 18-23 y) ultrasound application
Lesho26 Tuning fork; 128 Hz 52 soldiers with history and physical 6 clinicians Scintigraphy Marked exacerbation or reproduction
examination suggestive of tibial of shin pain in a localized area (3
stress fracture (mean age, 24.8 y; cm in diameter) with tuning fork
range, 19-43 y) application
Lowdon27 Ultrasound; 3-cm head; 0.75 59 patients with suspected stress 1 physiotherapist Roentgenogram; Pressure, bruising, and aching are the
MHz; intensity gradually fractures; age of patients not scintigraphy precise words needed to describe
increasing from 0 to 2.0 specified; 13 control subjects a positive ultrasound response
W/cm2
Moss and Ultrasound; 3-cm head; 0.75 35 consecutive patients (29 men) at- Not clearly specified; Roentgenogram; Very unpleasant sensation of intense
Mowat29 MHz; intensity gradually tending a sports injury clinic with his- ultrasound operator scintigraphy pressure or pain with ultrasound
increasing from 0 to 2.0 tory and clinical findings suggestive application
W/cm2 of a stress fracture of the leg (mean
age, 25 y; range, 16-51 y); 3 women
with rheumatoid arthritis (ages 25,
68, and 70) and possible stress
fracture related to osteoporosis and
treatment with corticosteroids were
also included; control group of 17 fit,
physically active subjects (mean age,
26 y; range, 16-45 y)
Nitz and Ultrasound; 2.0 to 3.0 W/cm2 54 male military basic trainees present- 2 therapists Roentgenogram Report of pain at the medial tibial
Scoville31 ing with knee pain (mean age not plateau with ultrasound
specified; range, 18-33 y) application
Romani et al36 Ultrasound; 5-cm2 head; 26 patients/subjects (12 men) recruited Not clearly specified; MRI Largest pain level reported on 100-
1 MHz; 0.0, 0.5, 1.0, 1.5, from 8 university and 2 recreational 1 radiologist mm blank visual analog scale
2.0, 2.5, and 2.9 W/cm2; activities with unilateral tibia pain with ultrasound application
30-s application (less than 2 wk duration) (mean age,
20.55 y; range, 18-31 y)
Wilder et al51 Tuning fork; 128 Hz, 256 Hz, 45 consecutive runners (42.2% men) 1 clinician Roentgenogram; MRI; Pain reported with tuning fork
512 Hz; 10-s application suspected of having stress fractures scintigraphy application on a standardized
(mean  SD age, 31.2  13.1 y) pain rating scale (0-4)
Abbreviations: MRI, magnetic resonance imaging; SD, standard deviation.
*Giladi et al18 tested both tibia in 53 subjects; Lowdon27 and Moss & Mowat29 tested 32 and 19 subjects with tibial stress fractures confirmed by scintigraphy
from their respective populations of 59 and 35.

for appropriate radiology also requires
that an adequate clinical examination
take place to further justify these some-
times expensive investigations. It has
been proposed that common and routine
nonradiological procedures used in phys-
ical therapy practice may be able to deter-
mine whether a stress fracture is present
independent of radiology; however, to
date, this has yet to be fully validated.
This systematic review and meta-analysis
evaluated the available evidence on the
applicability of clinical tests to diagnose
stress fractures of the lower limb. Re-
trieved studies included investigations
using therapeutic ultrasound and tuning
forks as clinical tests.
The application of ultrasound and the

766  |  september 2012  |  volume 42  |  number 9  |  journal of orthopaedic & sports physical therapy

42-09 Schneiders.indd 766 8/21/2012 4:14:06 PM


TABLE 3 QUADAS Scores for Each of the Included Studies

Item
Authors 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Total
Boam et al4 N, 0 N, 0 Y, 2 Y, 1 Y, 3 Y, 2 Y, 1 Y, 1 Y, 1 Y, 3 Y, 3 U, 0 Y, 1 Y, 1 19
Devereaux et al13 Y, 3 Y, 1 Y, 2 U, 0 Y, 3 Y, 2 Y, 1 Y, 1 Y, 1 U, 0 U, 0 Y, 3 Y, 1 Y, 1 19
Giladi et al18 Y, 3 U, 0 Y, 2 U, 0 Y, 3 Y, 2 Y, 1 U, 0 Y, 1 U, 0 U, 0 U, 0 Y, 1 Y, 1 14
Lesho26 Y, 3 Y, 1 Y, 2 N, 0 Y, 3 Y, 2 Y, 1 Y, 1 Y, 1 Y, 3 U, 0 Y, 3 Y, 1 Y, 1 22
Lowdon27 U, 0 U, 0 Y, 2 U, 0 N, 0 N, 0 Y, 1 Y, 1 Y, 1 U, 0 U, 0 U, 0 Y, 1 Y, 1 7
Moss and Mowat29 Y, 3 Y, 1 Y, 2 U, 0 N, 0 N, 0 Y, 1 Y, 1 Y, 1 Y, 3 U, 0 Y, 3 U, 0 U, 0 15
Nitz and Scoville31 Y, 3 Y, 1 N, 0 Y, 1 Y, 3 Y, 2 Y, 1 U, 0 Y, 1 Y, 3 U, 0 N, 0 Y, 1 Y, 1 17
Romani et al36 Y, 3 Y, 1 Y, 2 Y, 1 Y, 3 Y, 2 Y, 1 Y, 1 Y, 1 U, 0 Y, 3 Y, 3 Y, 1 Y, 1 23
Wilder et al51 Y, 3 Y, 1 Y, 2 N, 0 N, 0 N, 0 Y, 1 Y, 1 Y, 1 Y, 3 U, 0 U, 0 U, 0 U, 0 12
Percent agreement* 100 66.7 77.8 55.6 100 88.9 100 100 100 100 88.9 66.7 77.8 100 …
Abbreviations: N, no; U, unclear; Y, yes.
*Percent agreement between reviewers.

vibration produced by tuning forks over
superficial stress fracture sites have been
commonly reported to elicit pain,36,45 due
to the accelerated osteoclastic resorption
that occurs during the development of a
stress fracture, leading to periosteal dam-
age. This increased resorption results
in greater heat absorption at the stress
fracture site, which corresponds to an in-
crease in pain produced with ultrasound
application.36 The mechanism on which
tuning fork application evokes pain at
a stress fracture site differs from that of
ultrasound, as it is due to the vibratory
irritation of the damaged periosteum.45
An intact periosteum, or periosteum in
callus formation, is not affected by tun-
ing fork vibrations and does not absorb
sufficient energy from ultrasound to elicit
a painful response.36 This suggests that
the application of ultrasound or vibra-
tion from a tuning fork may cause pain
over stress fracture sites but not over ar-
eas without bony change or healed frac-
tures.1,26 Therefore, a time delay between
the application of the index and reference
tests, which was identified in some of the
articles, might have negatively affected
the results.
The results of the meta-analysis for
the ultrasound data indicated a pooled
sensitivity of 64% and specificity of 63%.
Although this suggests a low to moderate
performance of these measures for a posi-
tive test to identify a stress fracture and
a negative test to rule out a stress frac-
ture, it is important to note that the +LR
(2.09) and –LR (0.35), which are consid-
ered more clinically relevant measures
of a test’s diagnostic ability, were small
(TABLE 1). The 95% CI (0.7, 22.75) associ-
ated with the pooled DOR (6.2) was very
wide and included values considered to
be clinically useless (1.0), as well as those
with substantial clinical utility (greater
than 20). This indicates that these results
are too imprecise to draw any conclusion
about the usefulness of ultrasound to ac-
curately diagnose a stress fracture of the
lower limb in clinical practice. Addition-
ally, when studies were considered indi-
vidually, 4 of the ultrasound studies that
reported higher +LRs18,27,29,31 received rel-
atively low QUADAS scores, specifically,
due to poorly described participant selec-
tion and index testing. This suggests that
caution should be taken when interpret-
ing results from these studies due to po-
tential bias and validity issues. Moreover,
the 3 studies that scored relatively high
on the QUADAS tool4,13,36 (suggesting less
bias in the reporting of the results) had
very small +LRs.
We were unable to perform a meta-
analysis on the retrieved studies that
investigated tuning forks, and individual
results are therefore discussed. Lesho26
reported moderate to high sensitivity
(75%), moderate specificity (67%), and
a +LR of 2.3 for a 128-Hz tuning fork,
demonstrating a small but sometimes
important ability to identify a stress frac-
ture. The high score on the QUADAS tool
indicates that these results are less likely
to be subject to bias.
The study by Wilder et al51 scored
poorly on the QUADAS tool (12/26),
mainly due to unclear reporting of many
results, suggesting a high chance of bias.
These authors51 investigated the diagnos-
tic ability of 128-Hz, 256-Hz, and 512-Hz
tuning forks, comparing the results to 3
separate radiological reference tests.
The 256-Hz tuning fork had the high-
est reported sensitivity (92.3%, 90.0%,
and 77.7%) when compared to roent-
genograms, MRI, and bone scintigraphy,
respectively, usually indicating a high
probability of ruling out a stress frac-
ture if the test is negative. However, the
specificity values were low for the 256-Hz
tuning fork, suggesting difficulty diag-
nosing a stress fracture when the test is
positive. The other tuning fork frequen-
cies studied had lower sensitivities and
higher specificities, and the +LRs for all
frequencies were small to very small, sug-
gesting a limited ability of tuning forks to
diagnose stress fractures independent of

journal of orthopaedic & sports physical therapy  |  volume 42  |  number 9  |  september 2012  |  767

42-09 Schneiders.indd 767 8/21/2012 4:14:07 PM

 [ research report ]

TABLE 4 Summary of Diagnostic Accuracy Values

QUADAS
Sensitivity Specificity Score
Authors Application Site Tests (95% CI) (95% CI) +LR (95% CI) –LR (95% CI) DOR (95% CI) (26/26)
Boam et al4 Tibia Ultrasound, scintigraphy 43 (27, 61) 48 (34, 64) 1.07 (0.63, 1.72) 1.04 (0.67, 1.56) 1.08 (0.37, 2.46) 21
Devereaux et al13 Tibia, fibula Ultrasound, radiography, 53 (28, 78) 0 (0, 70) 1.41 (0.55, 2.81) 1.34 (0.33, 5.86) 2.03 (0.12, 7.27) 19
thermography,
scintigraphy
Giladi et al18 Tibia Ultrasound, radiography, 75 (60, 85) 67 (53, 79) 2.27 (1.57, 3.31) 0.37 (0.21, 0.57) 6.65 (2.74, 14.10) 14
scintigraphy
Lesho26 Tibia Tuning fork, scintigraphy 75 67 2.3 (1.10, 4.60) 0.37 (0.10, 0.71) ... 22
Lowdon27 Unclear, leg Ultrasound, radiography, 90 (67, 98) 75 (43, 93) 2.97 (1.62, 5.56) 0.17 (0.04, 0.39) 26.95 (3.61, 7
scintigraphy 114.60)
Moss and Mowat29 Unclear, leg Ultrasound, radiography, 91 (57, 100) 100 (60, 100) 3.67 (1.73, 8.18) 0.16 (0.02, 0.41) 171.20 (3.79, 15
scintigraphy 539.90)
Nitz and Scoville31 Medial tibial plateau Ultrasound, radiography 100 (88, 100) 80 (56, 94) 3.33 (2.14, 5.07) 0.06 (0.00, 0.17) ... 17
Romani et al36 Tibia Ultrasound, MRI 0 (0, 49) 100 (80, 100) 1.71 (0.33, 4.40) 0.91 (0.55, 1.18) 3.37 (0.02, 23
16.42)
Wilder et al51 Tibia, fibula, meta- Tuning fork, roent- ... ... ... ... ... 12
tarsal, calcaneus, genogram, MRI,
navicular, talus, scintigraphy
tarsal, phalanx
128 Hz ... ... R, 83; MRI, R, 38; MRI, R, 1.33 (0.92, 1.93); R, 0.44 (0.11, 1.74); ... ...
80; S, 71 50; S, 60 MRI, 1.60 (0.68, MRI, 0.40 (0.08,
3.77); S, 1.76 (0.58, 1.92); S, 0.49 (0.18,
5.40) 1.33)
256 Hz ... ... R, 92; MRI, R, 19; MRI, 20; R, 1.14 (0.91, 1.45); R, 0.40 (0.05, 3.40); ... ...
90; S, 78 S, 25 MRI, 1.08 (0.71, MRI, 0.60 (0.02,
1.63); S, 1.04 (0.56, 19.72); S, 0.89 (0.09,
1.92) 9.17
512 Hz ... ... R, 77; MRI, 50; R, 65; MRI, 83; R, 2.17 (1.24, 3.80); R, 0.36 (0.13, 1.00); ... ...
S, 35 S, 40 MRI, 3.00 (0.45, MRI, 0.60 (0.30,
19.90); S, 0.59 1.19); S, 1.62 (0.63,
(0.22, 1.54) 4.14)
Pooled (ultrasound … … 64 (55, 73) 63 (54, 71) 2.09 (1.10, 3.52) 0.35 (0.08, 0.86) 6.20 (0.70, …
studies only) 22.75)
Abbreviations: CI, confidence interval; DOR, diagnostic odds ratio; LR, likelihood ratio; MRI, magnetic resonance imaging; QUADAS, Quality Assessment of
Diagnostic Accuracy Studies; R, radiography; S, scintigraphy.

all 3 reference standards used. Addition-
ally, the –LRs for all tuning fork frequen-
cies were small to very small, with some
tests returning a –LR greater than 1. A
–LR greater than 1 means that a negative
test is more likely to occur in people with
the condition than in people without the
condition.1
Due to the heterogeneity of the study
results reported for tuning forks, the fact
that the authors did not report CIs for
sensitivity and specificity, and the poor
LRs for both studies, at best, these results
may not be conclusive and, at worst, tun-
ing forks may have limited clinical utili-
zation in the diagnosis of stress fractures
of the lower limb.
This systematic review only evaluated
2 of the tests commonly proposed to di-
agnose lower-limb stress fractures. Other
signs on clinical examination that have
been suggested to identify stress frac-
tures include localized bony tenderness,
pain on palpation over superficial stress
fracture sites,3,8,11,15,30,32 pain on loading or
impact of the affected bone,11,15,32 and pal-
pable thickening or swelling of the peri-
osteum over the subcutaneous bone.8,15,36
However, no studies evaluating these
clinical testing procedures met the inclu-
sion criteria for this review.
Limitations exist in most studies,
and these are also acknowledged in this
systematic review. All studies reviewed
examined athletes and/or military per-
sonnel, confirming that these popula-
tions are at risk for the development of
lower-limb stress fractures. Although
this makes the results applicable to these

768  |  september 2012  |  volume 42  |  number 9  |  journal of orthopaedic & sports physical therapy

42-09 Schneiders.indd 768 8/21/2012 4:14:08 PM

 specific populations, the results cannot be
generalized to other populations that also
have a high incidence of stress fractures,
including those with underlying patholo-
gies such as osteoporosis,17,39,48 rheuma-
toid arthritis,10,11,39 and cancer.22
The insertion of dummy values into
unpopulated cells to allow computation
of data in the meta-analysis might also
be considered a limitation of this review.
Although the use of diminutive values in
contingency tables is common practice in
logistic regression models,42 meta-analy-
sis of diagnostic studies is rare, and this
approach has not yet been fully validated,
despite the meta-analysis computer soft-
ware that does this by default. An inher-
ent limitation in any meta-analysis is that
it requires the elimination of publication
bias to obtain valid estimates21,38; howev-
er, this bias is suggested to be problem-
atic in all studies, whether randomized
trials, observational studies, or diagnostic
studies.
Two types of heterogeneity also con-
tribute to the limitations of this review.20
Clinical heterogeneity may exist due to
the mixed nature of the populations in
the contributing studies, whereas statisti-
cal heterogeneity may exist due to the in-
consistent results reported across studies,
which was clearly the case in this review.
Although MetaAnalyst is one of the few
programs that allow meta-analysis of di-
agnostic studies, it does not compute the
I2 statistic. Quantification of the degree of
heterogeneity of the studies in this meta-
analysis could not be done. Furthermore,
the extended time frame (1980-2009) of
the reviewed studies resulted in the use
of different statistics and methodologies
for the diagnostic studies. Therefore, the
results of this meta-analysis need to be
interpreted with caution. It is recom-
mended that future studies use the Stan-
dards for the Reporting of Diagnostic
Accuracy checklist to improve the accu-
racy and completeness of diagnostic ac-
curacy studies and their reporting.6 The
use of the Standards for the Reporting
of Diagnostic Accuracy checklist would
allow for improved reproducibility and
journal of orthopaedic & sports physical therapy  |  volume 42  |  number 9  |  september 2012  |  769
42-09 Schneiders.indd 769
homogeneity within the literature. An
additional limitation is due to the time
frame of the reviewed studies, as recent
advances in the technical capacity of ra-
diological imaging have likely contribut-
ed to the heterogeneity of the reference
data available for this review.
Inconsistencies also exist between
definitions of a stress reaction and stress
fracture in the literature, making compar-
ison of study results sometimes difficult.
An accepted definition for the 2 terms
would allow for more agreement be-
tween results and comparability between
studies, as it would be clear what patho-
logical stage of stress fracture was being
investigated. The use of 1 consistent gold
standard reference test would also allow
for more comparability between stud-
ies. There has been previous contention
within the literature as to whether MRI
or scintigraphy is the most appropriate
form of radiological imaging to identify
a stress fracture.5,18,26 However, it has re-
cently been suggested that single-photon
emission computed tomography may be
more effective than MRI and scintigra-
phy in the identification of lower-limb
stress fractures.9 Single-photon emission
computed tomography has been shown to
have higher sensitivity and accuracy than
scintigraphy alone in the evaluation and
grading of femoral neck stress fractures. 9
The use of single-photon emission com-
puted tomography should be investigated
in future studies to determine its ability
to be used as a gold standard measure
when diagnosing stress fractures.9
Recommendations for future research
can be proposed from the findings of this
systematic review. First, the test param-
eters utilized when applying ultrasound36
and tuning forks need to be fully stan-
dardized and reported6 to allow for test
reproducibility and conformity through-
out the literature. Second, it is suggested
that well-defined methods and larger
sample sizes be incorporated into future
research. Third, reference testing and in-
dex testing need to be undertaken within
an appropriate time frame to ensure
that both tests are completed without a
change in the stage of stress fracture pa-
thology. Additionally, methods reported
in the literature of obtaining patient re-
sponses to the application of ultrasound
or tuning forks vary, as these rely on the
patient’s reported response, which is not
dependable and is based on the individ-
ual’s interpretation of the rating method
used,18 the individual’s cooperation, and
other nonspecific factors such as pain tol-
erance and motivation. A common pain
or sensation rating scale should be uti-
lized in future studies to monitor patient
responses and to allow for more accurate
testing in the clinical setting.27,36
CONCLUSION
T
o our knowledge, this is the
first review to systematically evalu-
ate studies investigating the clinical
diagnosis of lower-limb stress fractures.
The strengths of this study are in its sys-
tematic search of the literature, the use
of meta-analytical methods to pool di-
agnostic data, and the use of a quality-
assurance tool (QUADAS) to evaluate all
included articles. The results of this study
do not support the specific use of ultra-
sound as a standalone diagnostic test for
lower-limb stress fractures. Additionally,
the literature supporting the use of tun-
ing forks needs to be interpreted with
caution, considering the limited number
of studies investigating this modality, the
differing results between different tun-
ing fork frequencies, and the reference
standard used. As the overall diagnostic
accuracy of the tests investigated is not
strong, based on the calculated LRs, it is
recommended that radiological imaging
should continue to be used for the confir-
mation and diagnosis of stress fractures
of the lower limb. t
KEY POINTS
FINDINGS: This review does not support
the use of therapeutic ultrasound or
tuning forks as a standalone test for the
diagnosis of lower-limb stress fractures.
IMPLICATIONS: Evaluation of the validity
of diagnostic tests provides clinicians
8/21/2012 4:14:10 PM
 [ research report
involved in the differential diagnosis of acetabular insufficiency fractures. Radiology.
lower-limb symptoms with evidence to 1985;157:15-17.
support or challenge the continuing use 11. Courtenay BG, Bowers DM. Stress fractures:
clinical features and investigation. Med J Aust.
of these tests.
1990;153:155-156.
CAUTION: Quantification of the degree of 12. Deeks JJ. Systematic reviews in health care:
heterogeneity of the studies included systematic reviews of evaluations of diagnostic
in the meta-analysis could not be done; and screening tests. BMJ. 2001;323:157-162.
13. Devereaux MD, Parr GR, Lachmann SM, Page-
therefore, the pooled results need to be
Thomas P, Hazleman BL. The diagnosis of stress
interpreted with caution. More high- fractures in athletes. JAMA. 1984;252:531-533.
quality studies, especially to determine 14. Edwards PH, Jr., Wright ML, Hartman JF. A
the diagnostic accuracy of tuning forks, practical approach for the differential diag-
nosis of chronic leg pain in the athlete. Am J
are required.
Sports Med. 2005;33:1241-1249. http://dx.doi.
org/10.1177/0363546505278305
ACKNOWLEDGEMENTS: Special thanks to Huang 15. Fredericson M, Bergman AG, Hoffman KL,
Ying for her assistance in the translation of the Dillingham MS. Tibial stress reaction in run-
ners. Correlation of clinical symptoms and
article published in Chinese.
scintigraphy with a new magnetic resonance
imaging grading system. Am J Sports Med.
1995;23:472-481.
REFERENCES 16. Fredericson M, Wun C. Differential diagnosis of
leg pain in the athlete. J Am Podiatr Med Assoc.
1. A kobeng AK. Understanding diagnostic tests 2003;93:321-324.
2: likelihood ratios, pre- and post-test prob- 17. Gibson MV. Evaluation and treatment of bone
abilities and their use in clinical practice. disease after fragility fracture. Geriatrics.
Acta Paediatr. 2007;96:487-491. http://dx.doi. 2008;63:21-30.
org/10.1111/j.1651-2227.2006.00179.x 18. Giladi M, Nili E, Ziv Y, Danon YL, Aharonson Z.
2. Alqarni AM, Schneiders AG, Hendrick PA. Comparison between radiography, bone scan,
Clinical tests to diagnose lumbar segmental and ultrasound in the diagnosis of stress frac-
instability: a systematic review. J Orthop Sports tures. Mil Med. 1984;149:459-461.
Phys Ther. 2011;41:130-140. http://dx.doi. 19. Greenhalgh T. How to read a paper. Papers
org/10.2519/jospt.2011.3457 that report diagnostic or screening tests. BMJ.
3. Bennell K, Brukner P. Preventing and managing 1997;315:540-543.
stress fractures in athletes. Phys Ther Sport. 20. Higgins JP, Thompson SG. Quantifying het-
2005;6:171-180. http://dx.doi.org/10.1016/j. erogeneity in a meta-analysis. Stat Med.
ptsp.2005.07.002 2002;21:1539-1558. http://dx.doi.org/10.1002/
4. Boam WD, Miser WF, Yuill SC, Delaplain CB, sim.1186
Gayle EL, MacDonald DC. Comparison of ultra- 21. Irwig L, Macaskill P, Glasziou P, Fahey M. Meta-
sound examination with bone scintiscan in the analytic methods for diagnostic test accuracy.
diagnosis of stress fractures. J Am Board Fam J Clin Epidemiol. 1995;48:119-130; discussion
Pract. 1996;9:414-417. 131-132.
5. Bolin D, Kemper A, Brolinson PG. Current 22. Ito K, Hori K, Terashima Y, Sekine M, Kura H. In-
concepts in the evaluation and management sufficiency fracture of the body of the calcaneus
of stress fractures. Curr Sports Med Rep. in elderly patients with osteoporosis: a report of
2005;4:295-300. two cases. Clin Orthop Relat Res. 2004:190-194.
6. Bossuyt PM, Reitsma JB, Bruns DE, et al. To- 23. Jaeschke R, Guyatt GH, Sackett DL. Users’
wards complete and accurate reporting of stud- guides to the medical literature. III. How to use
ies of diagnostic accuracy: the STARD initiative. an article about a diagnostic test. B. What are
Standards for Reporting of Diagnostic Accuracy. the results and will they help me in caring for
Clin Chem. 2003;49:1-6. my patients? The Evidence-Based Medicine
7. Breithaupt J. Zur Pathologie des menschlichen Working Group. JAMA. 1994;271:703-707.
Fußes. Med Zeitung. 1855;24:169-175. 24. Kaeding CC, Yu JR, Wright R, Amendola A,
8. Brukner P, Bennell K. Stress fractures in female Spindler KP. Management and return to
athletes. Diagnosis, management and rehabili- play of stress fractures. Clin J Sport Med.
tation. Sports Med. 1997;24:419-429. 2005;15:442-447.
9. Bryant LR, Song WS, Banks KP, Bui-Mansfield LT, 25. Kortebein PM, Kaufman KR, Basford JR, Stuart
Bradley YC. Comparison of planar scintigraphy MJ. Medial tibial stress syndrome. Med Sci
alone and with SPECT for the initial evalua- Sports Exerc. 2000;32:S27-33.
tion of femoral neck stress fracture. AJR Am J 26. Lesho EP. Can tuning forks replace bone scans
Roentgenol. 2008;191:1010-1015. http://dx.doi. for identification of tibial stress fractures? Mil
org/10.2214/AJR.07.3592 Med. 1997;162:802-803.
10. Cooper KL, Beabout JW, McLeod RA. Supra- 27. Lowdon A. Application of ultrasound to
770  |  september 2012  |  volume 42  |  number 9  |  journal of orthopaedic & sports physical therapy
42-09 Schneiders.indd 770
]
assess stress fractures. Physiotherapy.
1986;72:160-161.
28. Matheson GO, Clement DB, McKenzie DC,
Taunton JE, Lloyd-Smith DR, MacIntyre JG.
Stress fractures in athletes. A study of 320
cases. Am J Sports Med. 1987;15:46-58.
29. Moss A, Mowat AG. Ultrasonic assessment
of stress fractures. Br Med J (Clin Res Ed).
1983;286:1479-1480.
30. Murray SR, Reeder M, Ward T, Udermann BE.
Navicular stress fractures in identical twin
runners: high-risk fractures require structured
treatment. Phys Sportsmed. 2005;33:28-33.
http://dx.doi.org/10.3810/psm.2005.01.34
31. Nitz AJ, Scoville CR. Use of ultrasound in early
detection of stress fractures of the medial tibial
plateau. Mil Med. 1980;145:844-846.
32. Patel DR. Stress fractures: diagnosis and man-
agement in the primary care setting. Pediatr
Clin North Am. 2010;57:819-827. http://dx.doi.
org/10.1016/j.pcl.2010.03.004
33. Pepper M, Akuthota V, McCarty EC. The patho-
physiology of stress fractures. Clin Sports
Med. 2006;25:1-16. http://dx.doi.org/10.1016/j.
csm.2005.08.010
34. Portney L, Watkins M. Foundations of Clinical
Research: Applications to Practice. 3rd ed.
Upper Saddle River, NJ: Pearson/Prentice Hall;
2009.
35. Reeder MT, Dick BH, Atkins JK, Pribis AB,
Martinez JM. Stress fractures. Current con-
cepts of diagnosis and treatment. Sports Med.
1996;22:198-212.
36. Romani WA, Perrin DH, Dussault RG, Ball DW,
Kahler DM. Identification of tibial stress frac-
tures using therapeutic continuous ultrasound.
J Orthop Sports Phys Ther. 2000;30:444-452.
37. Schlesselman JJ, Stolley PD. Case-Control
Studies: Design, Conduct, Analysis. Oxford, UK:
Oxford University Press; 1982.
38. Song F, Khan KS, Dinnes J, Sutton AJ. Asym-
metric funnel plots and publication bias in
meta-analyses of diagnostic accuracy. Int J
Epidemiol. 2002;31:88-95.
39. Sourlas I, Papachristou G, Pilichou A, Gi-
annoudis PV, Efstathopoulos N, Nikolaou VS.
Proximal tibial stress fractures associated with
primary degenerative knee osteoarthritis. Am J
Orthop (Belle Mead NJ). 2009;38:120-124.
40. Sterling JC, Edelstein DW, Calvo RD, Webb
R, 2nd. Stress fractures in the athlete.
Diagnosis and management. Sports Med.
1992;14:336-346.
41. Strauch WB, Slomiany WP. Evaluating shin
pain in active patients. J Musculoskelet Med.
2008;25:138-148.
42. Suchower LJ, Copenhaver MD. Using logistic
regression to test for interaction in the presence
of zero cells. 10th Annual Northeast SAS Users
Group Conference; October 5-7, 1997; Baltimore,
MD.
43. Touliopolous S, Hershman EB. Lower leg pain.
Diagnosis and treatment of compartment syn-
dromes and other pain syndromes of the leg.
8/21/2012 4:14:10 PM
 Sports Med. 1999;27:193-204.
44. T uan K, Wu S, Sennett B. Stress fractures in ath-
letes: risk factors, diagnosis, and management.
Orthopedics. 2004;27:583-591; quiz 592-593.
45. van den Berg CD. The Reliability of the Tuning
Fork and Therapeutic Ultrasound to Determine
the Most Reliable Method of Fracture Diagnosis
[thesis]. Johannesburg, South Africa: Technikon
Witwatersrand; 2003.
46. Verma RB, Sherman O. Athletic stress fractures:
part II. The lower body. Part III. The upper body-
-with a section on the female athlete. Am J
binary, continuous and diagnostic data. BMC
Med Res Methodol. 2009;9:80. http://dx.doi.
org/10.1186/1471-2288-9-80
48. Wat SY, Seshadri N, Markose G, Balan K. Clinical
and scintigraphic evaluation of insufficiency
fractures in the elderly. Nucl Med Commun.
2007;28:179-185. http://dx.doi.org/10.1097/
MNM.0b013e328014690a
49. Whiting P, Harbord R, Kleijnen J. No role
for quality scores in systematic reviews
of diagnostic accuracy studies. BMC Med
Res Methodol. 2005;5:19. http://dx.doi.
for the quality assessment of studies of diag-
nostic accuracy included in systematic reviews.
BMC Med Res Methodol. 2003;3:25. http://
dx.doi.org/10.1186/1471-2288-3-25
51. Wilder RP, Vincent HK, Stewart J, Pack
C, Vincent KR. Clinical use of tuning
forks to identify running-related stress
fractures: a pilot study. Athl Train Sports
Health Care. 2009;1:12-18. http://dx.doi.
org/10.3928/19425864-20090101-10

@ MORE INFORMATION
Orthop (Belle Mead NJ). 2001;30:848-860. org/10.1186/1471-2288-5-19
47. Wallace BC, Schmid CH, Lau J, Trikalinos TA. 50. Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM,
Meta-Analyst: software for meta-analysis of Kleijnen J. The development of QUADAS: a tool WWW.JOSPT.ORG

PUBLISH Your Manuscript in a Journal With International Reach

JOSPT offers authors of accepted papers an international audience. The
Journal is currently distributed to the members of APTA’s Orthopaedic
and Sports Physical Therapy Sections and 23 orthopaedics, manual therapy,
and sports groups in 20 countries who provide online access as a member
benefit. As a result, the Journal is now distributed monthly to more than
32 000 individuals around the world who specialize in musculoskeletal and
sports-related rehabilitation, health, and wellness. In addition, JOSPT
reaches students and faculty, physical therapists and physicians at more
than 1,400 institutions in 56 countries. Please review our Information for and
Instructions to Authors at www.jospt.org and submit your manuscript for
peer review at http://mc.manuscriptcentral.com/jospt.

journal of orthopaedic & sports physical therapy  |  volume 42  |  number 9  |  september 2012  |  771

42-09 Schneiders.indd 771 8/21/2012 4:14:11 PM