This is an enhanced PDF from The Journal of Bone and Joint Surgery

The PDF of the article you requested follows this cover page.

The Accuracy of Diagnostic Imaging for the Assessment of Chronic

Osteomyelitis: A Systematic Review and Meta-Analysis
M.F. Termaat, P.G.H.M. Raijmakers, H.J. Scholten, F.C. Bakker, P. Patka and H.J.T.M. Haarman
J Bone Joint Surg Am. 2005;87:2464-2471. doi:10.2106/JBJS.D.02691

This information is current as of June 7, 2010

Supplementary Material http://www.ejbjs.org/cgi/content/full/87/11/2464/DC1

Reprints and Permissions
Publisher Information
Click here to order reprints or request permission to use material from this
article, or locate the article citation on jbjs.org and click on the [Reprints and
Permissions] link.
The Journal of Bone and Joint Surgery
20 Pickering Street, Needham, MA 02492-3157
www.jbjs.org
 2464
COPYRIGHT © 2005 BY THE JOURNAL OF BONE AND JOINT SURGERY, INCORPORATED

The Accuracy of Diagnostic Imaging

for the Assessment of Chronic
Osteomyelitis: A Systematic
Review and Meta-Analysis
BY M.F. TERMAAT, MD, P.G.H.M. RAIJMAKERS, MD, PHD, H.J. SCHOLTEN, MD,
F.C. BAKKER, MD, PHD, P. PATKA, MD, PHD, AND H.J.T.M. HAARMAN, MD, PHD
Investigation performed at the Departments of Surgery and Traumatology and Nuclear Medicine,
VU University Medical Center, Amsterdam, The Netherlands

Background: A variety of diagnostic imaging techniques is available for excluding or confirming chronic osteomyelitis. Un-
til now, an evidence-based algorithmic model for choosing the most suitable imaging technique has been lacking. The ob-
jective of this study was to determine the accuracy of current imaging modalities in the diagnosis of chronic osteomyelitis.
Methods: A systematic review and meta-analysis of the literature was conducted with a comprehensive search of the
MEDLINE, EMBASE, and Current Contents databases to identify clinical studies on chronic osteomyelitis that evaluated di-
agnostic imaging modalities. The value of each imaging technique was studied by determining its sensitivity and specificity
compared with the results of histological analysis, findings on culture, and clinical follow-up of more than six months.
Results: A total of twenty-three clinical studies in which the accuracy was described for radiography (two studies),
magnetic resonance imaging (five), computed tomography (one), bone scintigraphy (seven), leukocyte scintigraphy
(thirteen), gallium scintigraphy (one), combined bone and leukocyte scintigraphy (six), combined bone and gallium
scintigraphy (three), and fluorodeoxyglucose positron emission tomography (four) were included in the review. No
meta-analysis was performed with respect to computed tomography, gallium scintigraphy, and radiography. Pooled
sensitivity demonstrated that fluorodeoxyglucose positron emission tomography was the most sensitive technique,
with a sensitivity of 96% (95% confidence interval, 88% to 99%) compared with 82% (95% confidence interval, 70% to
89%) for bone scintigraphy, 61% (95% confidence interval, 43% to 76%) for leukocyte scintigraphy, 78% (95% confi-
dence interval, 72% to 83%) for combined bone and leukocyte scintigraphy, and 84% (95% confidence interval, 69%
to 92%) for magnetic resonance imaging. Pooled specificity demonstrated that bone scintigraphy had the lowest spec-
ificity, with a specificity of 25% (95% confidence interval, 16% to 36%) compared with 60% (95% confidence interval,
38% to 78%) for magnetic resonance imaging, 77% (95% confidence interval, 63% to 87%) for leukocyte scintigraphy,
84% (95% confidence interval, 75% to 90%) for combined bone and leukocyte scintigraphy, and 91% (95% confidence
interval, 81% to 95%) for fluorodeoxyglucose positron emission tomography. The sensitivity of leukocyte scintigraphy
in detecting chronic osteomyelitis in the peripheral skeleton was 84% (95% confidence interval, 72% to 91%) com-
pared with 21% (95% confidence interval, 11% to 38%) for its detection of chronic osteomyelitis in the axial skeleton.
The specificity of leukocyte scintigraphy in the axial skeleton was 60% (95% confidence interval, 39% to 78%) com-
pared with 80% (95% confidence interval, 61% to 91%) for the peripheral skeleton.
Conclusions: Fluorodeoxyglucose positron emission tomography has the highest diagnostic accuracy for confirming
or excluding the diagnosis of chronic osteomyelitis. Leukocyte scintigraphy has an appropriate diagnostic accuracy in
the peripheral skeleton, but fluorodeoxyglucose positron emission tomography is superior for detecting chronic osteo-
myelitis in the axial skeleton.
Level of Evidence: Diagnostic Level III. See Instructions to Authors for a complete description of levels of evidence.

I
nfections involving bone continue to be a common prob- small percentage of infections remain refractory to therapy,
lem in clinical practice. Despite the effectiveness of sur- leading to chronic osteomyelitis.
gical management and long-term antibiotic regimens, a Several classification systems for osteomyelitis based on
 2465
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG
VO L U M E 87-A · N U M B E R 11 · N O VE M B E R 2005
the etiology, or the physiology of the host, or on soft-tissue and
osseous defects1-4 have been published. Weiland et al. were the
first to establish the use of a period of six months to distinguish
between acute and chronic osteomyelitis5. Schauwecker stated
that osteomyelitis requiring more than one episode of treat-
ment and/or a persistent infection lasting more than six weeks
should be considered chronic6. In our opinion, the definition
used by Schauwecker provides a clear cutoff for acute and
chronic osteomyelitis, and the essence of the six-week period is
that it is more than three times longer than the ten-day period
during which bone necrosis occurs in acute osteomyelitis. In
contrast to acute purulent osteomyelitis, chronic osteomyelitis
is characterized by a low-grade presentation, with lymphocyte
and plasma cell infiltrates with variable degrees of necrosis and
osteosclerosis. This definition of chronic osteomyelitis in-
volves patients in whom osteomyelitis is persistent despite ade-
quate therapy or in whom the diagnosis of recurrent bone
infection is being considered. The diagnosis is difficult in these
patients as chronic bone infection alters the normal anatomy
and physiology of bone. These changes and the low-grade pre-
sentation require the utmost in diagnostic accuracy. Therefore,
this meta-analysis focused on patients with chronic osteomy-
elitis, as early diagnosis is the cornerstone of management. In
most patients, additional diagnostic imaging is essential to es-
tablish or rule out an active infection7,8.
Although a variety of imaging techniques is available, it
is often difficult to choose the most appropriate technique if
bone has been altered by previous infections, or bone regen-
eration, or if there are metallic implants7,8. The ideal imaging
technique should have a high sensitivity and specificity, and
numerous studies have been published concerning the accu-
racy of the various modalities in diagnosing chronic osteomy-
elitis. The objective of the present systematic review and
meta-analysis was to determine the accuracy of current clini-
cal imaging modalities in diagnosing chronic osteomyelitis.
Materials and Methods
Search Criteria
he diagnostic imaging techniques that were reviewed for
T the assessment of chronic osteomyelitis were radiography,
computed tomography, magnetic resonance imaging, leuko-
cyte scintigraphy, bone scintigraphy, gallium scintigraphy,
fluorodeoxyglucose positron emission tomography, and com-
bined techniques, such as combined bone and leukocyte scin-
tigraphy, and combined bone and gallium scintigraphy.
Search Strategy
A computer-aided search of the MEDLINE (from 1975), EM-
BASE (from 1980), and Current Contents databases was con-
ducted in October 2002 and was updated in July 2003 (see
Appendix). The search was restricted to primary studies that
were written in English and involved adults who were more
than nineteen years old. The search strategy of Mijnhout et al.
was used to develop a specific search strategy for each data-
base (Fig. 1)9. The reference lists of the identified studies and
relevant reviews were hand-searched for additional eligible
T H E ACC U R A C Y O F D I A G N O S T I C IM A G I N G F O R T H E AS S E S S M E N T
O F C H RO N I C O S T E O MYE L I T I S : A M E T A -A N A L Y S I S
studies, which were included in the final study selection.
Study Selection
Studies were eligible if they met the following inclusion crite-
ria. (1) The study was a clinical investigation that evaluated the
specified diagnostic index tests in patients with chronic osteo-
myelitis, which was defined as osteomyelitis requiring more
than one episode of treatment and/or a persistent infection
that had lasted for more than six weeks. (Studies that evaluated
septic or aseptic loosening of prostheses were not included.)
(2) The study group included ten patients or more. (3) The
study used a valid reference test (osteomyelitis was proven his-
tologically and/or bacteriologically, and/or there was a clinical
follow-up of more than six months in which no signs or symp-
toms of chronic infection were described). (4) Details provided
were sufficient to reconstruct a 2 × 2 contingency table express-
ing the results of the index tests by the disease status.
Exclusion criteria were (1) nonhuman studies and (2)
studies that included patients younger than nineteen years old
(because of the immaturity of the skeleton). If more than one
study by the same author was included, the absence of overlap
was carefully assessed by comparing the patient demograph-
ics or by contacting the author if these data were not provided.
The titles were screened for eligibility by one of the
reviewers (M.F.T.), and they were then processed for abstract
assessment. Two reviewers (M.F.T. and P.G.H.M.R.) indepen-
dently assessed the abstracts by consensus. All studies consid-
ered eligible or of dubious eligibility were retrieved, and the
final decision on inclusion was based on the full article.
Methodological Quality Assessment
The methodological quality of the studies was graded inde-
pendently by the same two reviewers (M.F.T. and P.G.H.M.R.),
using the criteria list for evaluating the internal and external
validity of diagnostic studies recommended by the Cochrane
Methods Group on Systematic Review of Screening and Diag-
nostic Tests (www.cochrane.org).
Using the above-mentioned inclusion and exclusion cri-
teria, two independent reviewers (M.F.T. and P.G.H.M.R.) se-
lected the remaining studies. Disagreement was resolved by
consensus or, if necessary, was independently settled by a third
reviewer. The internal validity criteria (A items) were used to
determine the methodological limitations of the studies, and
the external validity criteria (B and C items) were used for de-
scriptive purposes only.
Data Extraction
The two reviewers (M.F.T. and P.G.H.M.R.) independently ex-
tracted relevant data (patient characteristics, characteristics of
the index tests that were used, and outcome parameters) from
the included studies and tabulated the data using a standard-
ized form.
Statistical Methods
Qualitative Analysis
Levels of evidence were determined with use of the framework
 2466
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG
VO L U M E 87-A · N U M B E R 11 · N O VE M B E R 2005
Fig. 1
Flowchart for the conducted search strategy. MRI = magnetic resonance imaging, CT = computed tomography, and
PET-FDG = fluorodeoxyglucose positron emission tomography.
provided by The Journal of Bone and Joint Surgery (www.ejbjs.org).
Studies, however, were not excluded from the systematic review
on the basis of quality.
Quantitative Analysis (Meta-Analysis)
The sensitivity and specificity of the index tests were deter-
mined from the number of true-positive, false-positive, true-
negative, and false-negative results from the 2 × 2 contingency
table.
The statistical diagnostic heterogeneity of the sensitivity
and specificity per index test across studies was tested by the
chi-square test for independence with k-1 degrees of freedom
(k = the number of studies). In the case of heterogeneity, the
Spearman rank correlation coefficient ρ was used in order to
measure the correlation between sensitivity and specificity. A
ρ value of less than –0.40 suggests that the variation between
studies may be explained by different cutoff points, or diag-
nostic thresholds, on a summary receiver-operating character-
istic curve10,11.
T H E ACC U R A C Y O F D I A G N O S T I C IM A G I N G F O R T H E AS S E S S M E N T
O F C H RO N I C O S T E O MYE L I T I S : A M E T A -A N A L Y S I S
In the absence of a cutoff point difference, heterogene-
ity was explored by subgroup analysis on the basis of the a
priori hypothesis that the accuracy of a diagnostic test is in-
fluenced by the location of chronic osteomyelitis in the skele-
ton (i.e., the axial or central skeleton compared with the
peripheral skeleton), which was assessed with univariate re-
gression analyses12. Several authors have reported that labeled
leukocytes have a diminished sensitivity for imaging chronic
osteomyelitis in the axial skeleton6,12,13. The presence of he-
matopoietic bone marrow in the axial skeleton and alterations
in the microcirculation due to chronic infection have been
suggested to explain these so-called “cold lesions.” If the
included studies described the location in sufficient detail to
allow construction of 2 × 2 contingency tables for the axial
and peripheral skeleton, these data were extracted separately
on the data-extraction form. The influence of the internal (A-
criteria) and external (B-criteria) validity on diagnostic accu-
racy was assessed with univariate meta-regression analyses.
In the case of statistical heterogeneity of one or more of
 2467
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG
VO L U M E 87-A · N U M B E R 11 · N O VE M B E R 2005
the index tests, a random effect model was used, whereas, in
the case of statistical homogeneity, a fixed effect model was
used. These models calculate pooled estimates of the diag-
nostic accuracy. In addition, the models acknowledge the dif-
ference in precision by which sensitivity and specificity were
measured in each study (fixed effect model) and between
studies (random effect model)14. Sensitivity and specificity
were pooled independently and were weighted by the inverse
of the variance, with use of Meta-Test software (Lau J., Meta-
Test version 0.6; New England Medical Center, Boston, 1997)15-17.
Pooled estimates are presented with 95% confidence intervals.
The logit-transformed sensitivity, specificity, and corre-
sponding 95% confidence interval of the index tests were
compared with use of z-test statistics. A p value of <0.05 was
considered significant. All statistical analyses were performed
with the SPSS for Windows software program (version 11.0.1;
SPSS, Chicago, Illinois).
Results
Studies Included
he search strategy identified 1784 studies from MED-
T LINE, 1310 studies from EMBASE, and 290 studies from
the Current Contents database. The total of 3384 studies, in-
cluding duplicates, formed the source population for this re-
view. The studies retrieved from MEDLINE and EMBASE
showed overlap (23.1%), whereas no overlap was seen with
studies retrieved from the Current Contents database. Of the
initial 3384 studies, 3258 were excluded after review of the in-
formation provided in the title and abstract.
The full articles of the remaining 126 studies (fifty-six
studies from MEDLINE, sixty-six from EMBASE, and four
from the Current Contents database) were reviewed for eligi-
bility. Another fifty-six articles were extracted from the refer-
Fig. 2
Graphs showing the pooled estimates and corresponding confidence intervals of sensitivity and specificity for
all index tests. PET = positron emission tomography; BS = bone scintigraphy, LS = leukocyte scintigraphy, BS-
LS = bone and leukocyte scintigraphy, BS-Ga = bone and gallium scintigraphy, and MRI = magnetic resonance
imaging.
T H E ACC U R A C Y O F D I A G N O S T I C IM A G I N G F O R T H E AS S E S S M E N T
O F C H RO N I C O S T E O MYE L I T I S : A M E T A -A N A L Y S I S
ence lists of these articles, and thirty of them were reviewed
for inclusion. Ultimately, twenty-three studies (eighteen stud-
ies from MEDLINE, three from the reference lists, two from
EMBASE, and none from the Current Contents database)
were included in this review6,12,18-38. There was no disagreement
between the reviewers regarding final inclusion of the articles.
The reasons for exclusion were that the study had fewer than
ten patients (10%), an insufficient or unspecified reference
test (35%), no specified definition of chronic osteomyelitis
(33%), irreproducible 2 × 2 contingency tables (21%), or a
potential overlap of the patient population (1%). A list of the
excluded studies, with reasons for their rejection, is available
on request from the corresponding author (M.F.T.). In gen-
eral, no differentiation a priori was made for the etiology of
chronic osteomyelitis, although all included studies described
a patient population with a chronic bone infection due to
trauma and/or surgery and not to hematogenous contamina-
tion of bone.
Description of Study Characteristics
The twenty-three included studies described the diagnostic
accuracy of the imaging techniques for the detection of
chronic osteomyelitis with use of radiography (two stud-
ies)31,38, magnetic resonance imaging (five)20,21,29,31,35, computed
tomography (one)31, bone scintigraphy (seven)22,23,28,29,31,36,38,
leukocyte scintigraphy (thirteen)12,20,24,25,28-34,37,38, gallium scin-
tigraphy (one)38, combined bone and leukocyte scintigraphy
(six) 6,22,26,27,33,38, combined bone and gallium scintigraphy
(three)33,36,38, and fluorodeoxyglucose positron emission to-
mography (four)18-20,24. A total of 1269 diagnostic evaluations
for chronic osteomyelitis were performed in 687 patients
(54%) with disease and 582 patients (46%) without disease.
The characteristics of the studies and the diagnostic tests per-
 2468
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG
VO L U M E 87-A · N U M B E R 11 · N O VE M B E R 2005
TABLE I Diagnostic Accuracy of Radiography, Computed Tomography, and Gallium Scintigraphy for the Detection of
Chronic Osteomyelitis
No. of Patients No. of Patients
Technique with Disease without Disease
Radiography
Al-Sheikh et al.38 (1985) 10 12
Whalen et al.31 (1991) 9 2
Gallium scintigraphy
Al-Sheikh et al.38 (1985) 10 12
Computed tomography
Whalen et al.31 (1991) 6 2
formed are summarized by technique in the Appendix.
Critical Appraisal and Methodological Quality
The internal and external validity and the reproducibility of
the studies are presented in the Appendix.
Analysis
Qualitative Analysis
All studies had a level-of-evidence rating of III, which repre-
sents a nonconsecutive study or a study without consistently
applied reference standards. In nineteen of the twenty-three
studies, the reference test was not applied independently or
blindly. In four studies, an independent blind comparison was
performed, but it was done either in nonconsecutive patients
or in a narrow spectrum of patients, all of whom had under-
gone both the diagnostic test and the reference test.
Quantitative Analysis (Meta-Analysis)
Four studies (two involving radiography; one, gallium scintig-
raphy; and one, computed tomography) were excluded from
the analysis because they included too few patients (Table I).
Meta-analysis was performed for fluorodeoxyglucose
positron emission tomography, bone scintigraphy, leukocyte
scintigraphy, combined bone and leukocyte scintigraphy,
combined bone and gallium scintigraphy, and magnetic reso-
nance imaging. The sensitivity and specificity of the studies
according to the index test and the pooled estimates by ran-
dom effects analysis are described in the Appendix. The
pooled estimates of the diagnostic accuracy of all index tests
are given in Figure 2. Although included in the meta-analysis
of sensitivity, five studies were excluded from the analysis of
specificity because they did not provide information about the
proportion of patients without disease.
Fluorodeoxyglucose Positron Emission Tomography
The sensitivity and specificity of fluorodeoxyglucose positron
emission tomography were homogeneous (Qsens = 0.23, Qspec =
0.51: three degrees of freedom). The pooled sensitivity of
fluorodeoxyglucose positron emission tomography was sig-
nificantly higher than that of the other tests (p < 0.05). The
specificity of fluorodeoxyglucose positron emission tomogra-
phy was significantly higher than that of leukocyte scintig-
T H E ACC U R A C Y O F D I A G N O S T I C IM A G I N G F O R T H E AS S E S S M E N T
O F C H RO N I C O S T E O MYE L I T I S : A M E T A -A N A L Y S I S
Sensitivity Specificity
(95% Confidence Interval) (95% Confidence Interval)
0.60 (0.28-0.86) 0.67 (0.36-0.89)
0.78 (0.40-0.96) 1.00 (0.16-1.00)
0.80 (0.44-0.96) 0.42 (0.17-0.71)
0.67 (0.24-0.94) 0.50 (0.03-0.97)
raphy (p = 0.033), bone scintigraphy (p = 0.0001), and
magnetic resonance imaging (p = 0.0011), but it was not
found to be significantly different from that of combined
bone and leukocyte scintigraphy (p = 0.17) and combined
bone and gallium scintigraphy (p = 0.10).
Bone Scintigraphy
The sensitivity and specificity of bone scintigraphy were ho-
mogeneous (Qsens = 9.34: seven degrees of freedom, Qspec =
6.86: six degrees of freedom). The sensitivity of bone scintig-
raphy was significantly higher than that of leukocyte scintigra-
phy (p = 0.027), but it was not different from that of the
combined techniques or magnetic resonance imaging.
Leukocyte Scintigraphy
The sensitivity and specificity of leukocyte scintigraphy were
heterogeneous (Qsens = 54.46: twelve degrees of freedom, Qspec =
20.39: ten degrees of freedom). The Spearman correlation fac-
tor was –0.032, indicating that the heterogeneity could not be
explained by different cutoff points. The meta-regression re-
vealed a significant association between the variance in the ac-
curacy of leukocyte scintigraphy with the location of chronic
osteomyelitis in the skeleton (axial versus peripheral skeleton)
(p < 0.05), but not with any of the other covariates.
The sensitivity of leukocyte scintigraphy was not signifi-
cantly different from that of combined bone and gallium scin-
tigraphy (p = 0.78) and was significantly lower than that of the
other tests (p < 0.05). No significant difference in specificity
was detected between leukocyte scintigraphy and combined
bone and leukocyte scintigraphy (p = 0.32), combined bone
and gallium scintigraphy (p = 0.16), or magnetic resonance
imaging (p = 0.94).
Subgroup Analysis of Leukocyte Scintigraphy
The sensitivity and specificity of leukocyte scintigraphy were
homogeneous for the axial skeleton (Qsens, axial = 7.65: five de-
grees of freedom, Qspec, axial = 7.96: four degrees of freedom). In
contrast, the sensitivity was homogeneous but the specificity
was heterogeneous for detecting chronic osteomyelitis in the
peripheral skeleton (Qsens, peripheral = 10.99, Qspec, peripheral = 15.39: six
degrees of freedom). The sensitivity of leukocyte scintigraphy
in the axial skeleton was significantly lower than that in the
 2469
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG
VO L U M E 87-A · N U M B E R 11 · N O VE M B E R 2005
peripheral skeleton (p = 0.0010), whereas the difference in
specificity was not significant. The sensitivity of leukocyte
scintigraphy in the axial skeleton was significantly lower than
that of the other techniques, whereas the sensitivity in the pe-
ripheral skeleton was not significantly different from that of
the other techniques except for fluorodeoxyglucose positron
emission tomography.
Combined Bone and Leukocyte Scintigraphy
The sensitivity of combined bone and leukocyte scintigraphy
was homogeneous (Qsens = 8.90: five degrees of freedom),
whereas its specificity was heterogeneous (Qspec = 15.26: five
degrees of freedom). Combined bone and leukocyte scintigra-
phy had a significantly higher sensitivity than leukocyte scin-
tigraphy (p = 0.04) and a significantly higher specificity than
bone scintigraphy (p = 0.001). The differences in specificity
for leukocyte scintigraphy and sensitivity for bone scintigra-
phy were not significant.
Combined Bone and Gallium Scintigraphy
The sensitivity of combined bone and gallium scintigraphy
was heterogeneous (Qsens = 8.04: two degrees of freedom), and
its specificity was homogeneous (Qspec = 3.27: two degrees of
freedom). Combined bone and gallium scintigraphy had a
significantly higher specificity (p = 0.0006), but not sensitivity
(p = 0.07), than bone scintigraphy alone. The sensitivity and
specificity of combined bone and gallium scintigraphy did not
significantly differ from that of combined bone and leukocyte
scintigraphy.
Magnetic Resonance Imaging
The sensitivity and specificity of magnetic resonance imag-
ing were homogeneous (Qsens = 4.62: four degrees of free-
dom, Qspec = 0.02: two degrees of freedom). The sensitivity of
magnetic resonance imaging for chronic osteomyelitis in the
peripheral skeleton was not different from that of leukocyte
scintigraphy or combined bone and gallium scintigraphy.
Discussion
his meta-analysis showed that fluorodeoxyglucose posi-
T tron emission tomography is the most sensitive technique
for detecting chronic osteomyelitis and that it has a greater
specificity than leukocyte scintigraphy, bone scintigraphy, or
magnetic resonance imaging. Fluorodeoxyglucose positron
emission tomography was better than leukocyte scintigraphy
for imaging chronic osteomyelitis in the axial skeleton.
Several authors have reported a loss of sensitivity of
leukocyte scintigraphy when imaging chronic osteomyelitis
in the axial skeleton6,12,13. The nonspecific uptake of labeled
leukocytes has been suggested to be due to the presence of
hematopoietic bone marrow in the axial skeleton. The de-
creased uptake of labeled cells in infections of the axial skel-
eton, so-called “cold lesions,” is poorly understood, but
microthrombotic occlusion and inflammatory compression
of the blood vessels may prevent labeled cells from reaching
the site of infection. Indeed, the observed variance in diag-
T H E ACC U R A C Y O F D I A G N O S T I C IM A G I N G F O R T H E AS S E S S M E N T
O F C H RO N I C O S T E O MYE L I T I S : A M E T A -A N A L Y S I S
nostic accuracy of leukocyte scintigraphy in this meta-analysis
could be explained by the location of the chronic osteomyeli-
tis. Subgroup analyses demonstrated that the pooled sensitiv-
ity of leukocyte scintigraphy decreased significantly (from
84% to 21%) for chronic osteomyelitis of the peripheral and
axial skeleton, respectively. Indeed, leukocyte scintigraphy
for chronic osteomyelitis in the axial skeleton was the least
sensitive of the imaging techniques. Thus, while leukocyte
scintigraphy is an accurate technique for the detection of
chronic osteomyelitis in the peripheral skeleton, it is an inap-
propriate technique for diagnosing chronic osteomyelitis in
the axial skeleton. In the studies analyzed, white blood cells
were labeled with technetium99m or indium111, but there
was no evidence that the labeling technique affected the diag-
nostic accuracy of the method.
Bone scintigraphy has been suggested to be a sensitive
technique to screen for bone alterations caused by chronic os-
teomyelitis. However, in this analysis, the sensitivity of bone
scintigraphy for chronic osteomyelitis was not significantly
different from that of leukocyte scintigraphy, and it was the
least specific of the techniques investigated. This low specific-
ity has led to the suggestion that bone scintigraphy should be
combined with other techniques, such as leukocyte scintigra-
phy or gallium scintigraphy. Although the specificity of the
combined techniques was significantly higher, it was still not
significantly different from that of leukocyte scintigraphy as a
single technique. Hence, leukocyte scintigraphy seems more
appropriate for evaluating chronic osteomyelitis, at least in the
peripheral skeleton.
Some radiographic techniques available for diagnosing
chronic osteomyelitis, such as radiography, computed tomog-
raphy, and magnetic resonance imaging, were poorly repre-
sented in this meta-analysis. Although there were several studies
of magnetic resonance imaging, there were few that described
the accuracy of radiography and computed tomography. Mag-
netic resonance imaging is very sensitive for detecting bone al-
terations, such as those that occur in chronic osteomyelitis. In
general, magnetic resonance imaging can show sites with tissue
edema and increased regional perfusion. These observations
can also be observed for a long time after bone surgery. There-
fore, magnetic resonance imaging lacks specificity unless clear
morphological signs for osteomyelitis are observed, and it is of
limited value if there are metal implants. Thus, other diag-
nostic procedures are often needed21. The sensitivity of mag-
netic resonance imaging was comparable with that of leukocyte
scintigraphy in the peripheral skeleton, although the specificity
was lower. For chronic osteomyelitis located in the axial skele-
ton, the diagnostic accuracy of fluorodeoxyglucose positron
emission tomography was significantly higher than that of
magnetic resonance imaging.
We assessed the methodological quality of the primary
studies because it affects the interpretation of results14,39.
Rather than excluding studies of poor quality, we chose to an-
alyze each component of the study validity in a statistical re-
gression model to determine the effect of design flaws on the
estimates of diagnostic accuracy. However, differences in accu-
 2470
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG
VO L U M E 87-A · N U M B E R 11 · N O VE M B E R 2005
racy could not be explained by regression analysis of the
methodological criteria.
In general, the methodological quality of a diagnostic
test is judged against a valid reference test with an indepen-
dent selection of subjects for the index and reference tests, and
by whether these are assessed independently of each other. In
this meta-analysis, a valid reference test was an inclusion crite-
rion, but one-third of the studies did not meet this criterion.
While most studies described the independent selection of pa-
tients for the reference test and an independent comparison of
the index test, 13% and 17% of the studies, respectively, did
not. Furthermore, few of the primary studies described the re-
producibility (interobserver variance) of the imaging tech-
nique investigated, which is a major study limitation because
diagnostic results may be hampered by large interobserver
variability.
Systematic reviews are hampered by publication bias,
i.e., the preferential publication of studies with significant
positive results rather than those with negative results14. How-
ever, Olson et al. found that, among submitted manuscripts,
no significant difference was detected with respect to the pub-
lication rates of those with positive results compared with
those with negative results40. Advantages of systematic reviews
and meta-analyses are an increase in statistical power and in
the ability to assess sources of heterogeneity when there is no
consistency in the reported accuracy across studies and to
provide overall estimates of diagnostic accuracy. With this
approach, a ratio for clinical decision-making on how to inter-
pret and generalize these results can be made on the aggregate
of pertinent knowledge rather than on the basis of the results
of a single study or of personal experience.
This meta-analysis demonstrated that fluorodeoxyglu-
cose positron emission tomography has the highest accuracy
for confirming or excluding the diagnosis of chronic osteomy-
elitis. Hence, on the basis of the current evidence, fluorodeox-
yglucose positron emission tomography might be used as the
first choice in a patient with a suspected chronic bone infec-
tion. Because of the limited availability of positron emission
tomography systems, leukocyte scintigraphy can be used with
satisfactory diagnostic accuracy for detecting chronic osteo-
References
1. Gordon L, Chiu EJ. Treatment of infected non-unions and segmental defects
of the tibia with staged microvascular muscle transplantation and bone-grafting.
J Bone Joint Surg Am. 1988;70:377-86.
2. Waldvogel FA, Medoff G, Swartz MN. Osteomyelitis: a review of clinical fea-
tures, therapeutic considerations and unusual aspects. 3. Osteomyelitis associ-
ated with vascular insufficiency. N Engl J Med. 1970;282:316-22.
3. Kelly PJ. Infected nonunion of the femur and tibia. Orthop Clin North Am.
1984;15:481-90.
4. May JW Jr, Jupiter JB, Weiland AJ, Byrd HS. Clinical classification of post-
traumatic tibial osteomyelitis. J Bone Joint Surg Am. 1989;71:1422-8.
5. Weiland AJ, Moore JR, Daniel RK. The efficacy of free tissue transfer in the
treatment of osteomyelitis. J Bone Joint Surg Am. 1984;66:181-93.
6. Schauwecker DS. Osteomyelitis: diagnosis with In-111-labeled leukocytes. Ra-
diology. 1989;171:141-6.
7. Kaim AH, Gross T, von Schulthess GK. Imaging of chronic posttraumatic os-
teomyelitis. Eur Radiol. 2002;12:1193-202.
T H E ACC U R A C Y O F D I A G N O S T I C IM A G I N G F O R T H E AS S E S S M E N T
O F C H RO N I C O S T E O MYE L I T I S : A M E T A -A N A L Y S I S
myelitis in the peripheral skeleton. Prospective, randomized
clinical trials are needed to confirm the findings of these meta-
analyses and to investigate the capability of fluorodeoxyglu-
cose positron emission tomography technology to quantify
the rate of infection and thus to evaluate, in an early phase, the
response to treatment of chronic osteomyelitis.
Appendix
The search string used to identify articles, the character-
istics of the included articles, the methodological quality
assessment criteria, and figures showing the diagnostic accu-
racy of each of the specific modalities are available with the
electronic versions of this article, on our web site at jbjs.org
(go to the article citation and click on “Supplementary Mate-
rial”) and on our quarterly CD-ROM (call our subscription
department, at 781-449-9780, to order the CD-ROM). 
M.F. Termaat, MD
P.G.H.M. Raijmakers, MD, PhD
H.J. Scholten, MD
F.C. Bakker, MD, PhD
P. Patka, MD, PhD
H.J.T.M. Haarman, MD, PhD
Departments of Surgery and Traumatology (M.F.T., H.J.S., F.C.B., P.P.,
and H.J.T.M.H.) and Nuclear Medicine (P.G.H.M.R.), VU University
Medical Center, P.O. Box 7057, 1007 MB Amsterdam, The Netherlands.
E-mail address for M.F. Termaat: mf.termaat@vumc.nl. E-mail address
for P.G.H.M. Raijmakers: p.raijmakers@vumc.nl. E-mail address for H.J.
Scholten: h_scholten@hotmail.com. E-mail address for F.C. Bakker:
fc.bakker@vumc.nl. E-mail address for P. Patka: p.patka@vumc.nl.
E-mail address for H.J.T.M. Haarman: hjtm.haarman@vumc.nl
In support of their research or preparation of this manuscript, one or
more of the authors received grants or outside funding from The Associ-
ation of University Hospitals (VAZ) and Health Care Research Board
(CVZ). None of the authors received payments or other benefits or a
commitment or agreement to provide such benefits from a commercial
entity. No commercial entity paid or directed, or agreed to pay or direct,
any benefits to any research fund, foundation, educational institution, or
other charitable or nonprofit organization with which the authors are af-
filiated or associated.
doi:10.2106/JBJS.D.02691
8. Gross T, Kaim AH, Regazzoni P, Widmer AF. Current concepts in posttraumatic
osteomyelitis: a diagnostic challenge with new imaging options. J Trauma.
2002;52:1210-9.
9. Mijnhout GS, Hoekstra OS, van Tulder MW, Teule GJ, Deville WL. Systematic
review of the diagnostic accuracy of (18)F-fluorodeoxyglucose positron emission
tomography in melanoma patients. Cancer. 2001;91:1530-42.
10. Moses LE, Shapiro D, Littenberg B. Combining independent studies of a di-
agnostic test into a summary ROC curve: data-analytic approaches and some ad-
ditional considerations. Stat Med. 1993;12:1293-316.
11. Deville WL, Buntinx F, Bouter LM, Montori VM, de Vet HC, van der Windt DA,
Bezemer PD. Conducting systematic reviews of diagnostic studies: didactic guide-
lines. BMC Med Res Methodol. 2002;2:9.
12. Wukich DK, Abreu SH, Callaghan JJ, Van Nostrand D, Savory CG, Eggli DF,
Garcia JE, Berrey BH. Diagnosis of infection by preoperative scintigraphy with
indium-labeled white blood cells. J Bone Joint Surg Am. 1987;69:1353-60.
13. Gratz S, Braun HG, Behr TM, Meller J, Herrmann A, Conrad M, Rathmann D,
 2471
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG
VO L U M E 87-A · N U M B E R 11 · N O VE M B E R 2005
Bertagnoli R, Willert HG, Becker W. Photopenia in chronic vertebral osteomyelitis
with technetium-99m-antigranulocyte antibody (BW 250/183). J Nucl Med. 1997;
38:211-6.
14. Irwig L, Macaskill P, Glasziou P, Fahey M. Meta-analytic methods for diagnos-
tic test accuracy. J Clin Epidemiol. 1995;48:119-30.
15. Littenberg B, Moses LE. Estimating diagnostic accuracy from multiple conflict-
ing reports: a new meta-analytic method. Med Decis Making. 1993;13:313-21.
16. Moses LE, Shapiro D, Littenberg B. Combining independent studies of a di-
agnostic test into a summary ROC curve: data-analytic approaches and some ad-
ditional considerations. Stat Med. 1993;12:1293-316.
17. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials.
1986;7:177-88.
18. Schiesser M, Stumpe KD, Trentz O, Kossmann T, Von Schulthess GK. Detec-
tion of metallic implant-associated infections with FDG PET in patients with
trauma: correlation with microbiologic results. Radiology. 2003;226:391-8.
19. De Winter F, Van de Wiele C, Vogelaers D, de Smet K, Verdonk R, Dierckx RA.
Fluorine-18 fluorodeoxyglucose-position emission tomography: a highly accurate
imaging modality for the diagnosis of chronic musculoskeletal infections. J Bone
Joint Surg Am. 2001;83:651-60.
20. Meller J, Koster G, Liersch T, Siefker U, Lehmann K, Meyer I, Schreiber K, Alt-
envoerde G, Becker W. Chronic bacterial osteomyelitis: prospective comparison
of (18)F-FDG imaging with a dual-head coincidence camera and (111)In-labelled
autologous leucocyte scintigraphy. Eur J Nucl Med Mol Imaging. 2002;29:53-60.
21. Kaim A, Ledermann HP, Bongartz G, Messmer P, Muller-Brand J, Steinbrich W.
Chronic post-traumatic osteomyelitis of the lower extremity: comparison of mag-
netic resonance imaging and combined bone scintigraphy/immunoscintigraphy
with radiolabelled monoclonal antigranulocyte antibodies. Skeletal Radiol. 2000;
29:378-86.
22. Kaim A, Maurer T, Ochsner P, Jundt G, Kirsch E, Mueller-Brand J. Chronic
complicated osteomyelitis of the appendicular skeleton: diagnosis with techne-
tium-99m labelled monoclonal antigranulocyte antibody-immunoscintigraphy. Eur
J Nucl Med. 1997;24:732-8.
23. Nijhof MW, Oyen WJ, van Kampen A, Claessens RA, van der Meer JW,
Corstens FH. Evaluation of infections of the locomotor system with indium-111-
labeled human IgG scintigraphy. J Nucl Med. 1997;38:1300-5.
24. Guhlmann A, Brecht-Krauss D, Suger G, Glatting G, Kotzerke J, Kinzl L,
Reske SN. Fluorine-18-FDG PET and technetium-99m antigranulocyte antibody
scintigraphy in chronic osteomyelitis. J Nucl Med. 1998;39:2145-52.
25. Devillers A, Moisan A, Jean S, Arvieux C, Bourguet P. Technetium-99m hexa-
methylpropylene amine oxime leucocyte scintigraphy for the diagnosis of bone
and joint infections: a retrospective study in 116 patients. Eur J Nucl Med.
1995;22:302-7.
26. Nepola JV, Seabold JE, Marsh JL, Kirchner PT, el-Khoury GY. Diagnosis of in-
fection in ununited fractures. Combined imaging with indium-111-labeled leuko-
cytes and technetium-99m methylene diphosphonate. J Bone Joint Surg Am.
1993;75:1816-22.
27. Krznaric E, Roo MD, Verbruggen A, Stuyck J, Mortelmans L. Chronic osteomy-
T H E ACC U R A C Y O F D I A G N O S T I C IM A G I N G F O R T H E AS S E S S M E N T
O F C H RO N I C O S T E O MYE L I T I S : A M E T A -A N A L Y S I S
elitis: diagnosis with technetium-99m-d, l-hexamethylpropylene amine oxime la-
belled leucocytes. Eur J Nucl Med. 1996;23:792-7.
28. Sciuk J, Brandau W, Vollet B, Stucker R, Erlemann R, Bartenstein P, Peters
PE, Schober O. Comparison of technetium 99m polyclonal human immunoglobu-
lin and technetium 99m monoclonal antibodies for imaging chronic osteomyelitis.
First clinical results. Eur J Nucl Med. 1991;18:401-7.
29. Williamson MR, Quenzer RW, Rosenberg RD, Meholic AJ, Eisenberg B, Espi-
nosa MC, Hartshorne MF. Osteomyelitis: sensitivity of 0.064 T MRI, three-phase
bone scanning and indium scanning with biopsy proof. Magn Reson Imaging.
1991;9:945-8.
30. Flivik G, Sloth M, Rydholm U, Herrlin K, Lidgren L. Technetium-99m-
nanocolloid scintigraphy in orthopedic infections: a comparison with indium-
111-labeled leukocytes. J Nucl Med. 1993;34:1646-50.
31. Whalen JL, Brown ML, McLeod R, Fitzgerald RH Jr. Limitations of indium leu-
kocyte imaging for the diagnosis of spine infections. Spine. 1991;16:193-7.
32. el Esper I, Dacquet V, Paillard J, Bascoulergue G, Tahon MM, Fonroget J.
99Tcm-HMPAO-labelled leucocyte scintigraphy in suspected chronic osteomyeli-
tis related to an orthopaedic device: clinical usefulness. Nucl Med Commun.
1992;13:799-805.
33. Seabold JE, Nepola JV, Conrad GR, Marsh JL, Montgomery WJ, Bricker JA,
Kirchner PT. Detection of osteomyelitis at fracture nonunion sites: comparison
of two scintigraphic methods. AJR Am J Roentgenol. 1989;152:1021-7.
34. Palestro CJ, Kim CK, Swyer AJ, Vallabhajosula S, Goldsmith SJ. Radionuclide
diagnosis of vertebral osteomyelitis: indium-111-leukocyte and technetium-
99m-methylene diphosphonate bone scintigraphy. J Nucl Med. 1991;32:
1861-5.
35. Mason MD, Zlatkin MB, Esterhai JL, Dalinka MK, Velchik MG, Kressel HY.
Chronic complicated osteomyelitis of the lower extremity: evaluation with MR
imaging. Radiology. 1989;173:355-9.
36. Esterhai J, Alavi A, Mandell GA, Brown J. Sequential technetium-99m/
gallium-67 scintigraphic evaluation of subclinical osteomyelitis complicating frac-
ture nonunion. J Orthop Res. 1985;3:219-25.
37. Esterhai JL Jr, Goll SR, McCarthy KE, Velchik M, Alavi A, Brighton CT, Heppen-
stall RB. Indium-111 leukocyte scintigraphic detection of subclinical osteomyelitis
complicating delayed and nonunion long bone fractures: a prospective study. J Or-
thop Res. 1987;5:1-6.
38. Al-Sheikh W, Sfakianakis GN, Mnaymneh W, Hourani M, Heal A, Duncan RC,
Burnett A, Ashkar FS, Serafini AN. Subacute and chronic bone infections: diagno-
sis using In-111, Ga-67 and Tc-99m MDP bone scintigraphy, and radiography. Ra-
diology. 1985;155:501-6.
39. Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP, Irwig LM,
Lijmer JG, Moher D, Rennie D, de Vet HC; Standards for Reporting of Diagnostic
Accuracy Group. Towards complete and accurate reporting of studies of diagnos-
tic accuracy: the STARD initiative. BMJ. 2003;326:41-4.
40. Olson CM, Rennie D, Cook D, Dickersin K, Flanagin A, Hogan JW, Zhu Q, Reil-
ing J, Pace B. Publication bias in editorial decision making. JAMA.
2002;287:2825-8.