14
Imaging of Acute Hematogenous Osteomyelitis and Septic Arthritis in Children and Adults
John Y. Kim and Diego Jaramillo
I. What are the clinical findings that raise the suspicion for acute hematogenous osteomyelitis and septic arthritis to direct further imaging?
II. What is the diagnostic performance of the different imaging studies in acute hematogenous osteomyelitis and septic arthritis?
III. What is the natural history of osteomyelitis and septic arthritis, and what are the roles of medical therapy versus surgical treatment?
IV. Is there a role for repeat imaging in the management?
V. What is the diagnostic performance of imaging of osteomyelitis and septic arthritis in the adult?
VI. What are the roles of the different imaging modalities in the evalu- ation of acute osteomyelitis and septic orthritis?
260
Issues
䊏
The clinical presentation of acute osteomyelitis and septic arthritis can be nonspecific and sometimes confusing (moderate evidence).
䊏
When signs and symptoms cannot be localized, bone scintigraphy is preferred over magnetic resonance imaging (MRI) (moderate evidence).
䊏
When signs and symptoms can be localized, MRI is preferred (mod- erate to limited evidence).
䊏
Ultrasound is the preferred imaging modality for evaluating joint effusions of the hip (moderate evidence).
䊏 Magnetic resonance imaging is highly sensitive for the detection of
osteomyelitis and its complications (abscess), but incurs added cost (moderate evidence).
䊏
No data were found in the medical literature that evaluate the cost- effectiveness of the different imaging modalities in the evaluation of hematogenous osteomyelitis and septic joint (limited evidence).
䊏
Overall, MRI is the imaging modality of choice to evaluate for oste- omyelitis and septic arthritis in the adult population, including the diabetic patient and intravenous drug users. The ability to localize
Key Points
Definition and Pathophysiology
Osteomyelitis is an infection of bone and bone marrow. Routes of infection include hematogenous spread, spread by contiguity, and direct infection by a penetrating wound (1). Hematogenous spread is the most common route in children, usually seeding the metaphyses of long bones due to sluggish blood flow patterns in this region (2,3). It arises in the setting of bacteremia. In children, the capillaries in the metaphyses are the terminal branches of the nutrient artery. The capillaries form loops that end in large venous sinusoids where there is decreased blood flow. The inflammatory response to infection leads to increased intraosseous pressure and stasis of blood flow, causing thrombosis and eventual bone necrosis (4). In children less than 18 months of age, transphyseal vessels allow metaphyseal infec- tions to cross the physis and infect the epiphyses and joints. The most common bones affected by acute hematogenous osteomyelitis (AHO) are the tibia and femur (3); the most common organism is Staphylococcus aureus.
Acute septic arthritis is a bacterial infection of a joint. Most cases arise from hematogenous spread or contiguous spread from adjacent osteomyelitis in the metaphysis or epiphysis (5–7). The most common organism is S. aureus (3). The prognosis worsens with increasing delay of treatment due to lytic enzymes that destroy the articular and epiphyseal cartilage. In addition, increased pressure within the joint capsule reduces blood flow to the epiphyses. This can lead to long-term disability result- ing from growth disturbances, dislocations, and malalignment (8,9).
There is evidence that acute osteomyelitis and septic arthritis are a spec- trum of the same disease process (moderate evidence) (10). This hypothesis argues for a similar clinical approach and treatment for these two diseases.
The pattern of hematogenous spread of osteomyelitis and septic arthri- tis in the adult is different from the pediatric population. The unique vas- cular supply in the metaphysis normally seen in children is no longer present in adults, and most hematogenous infections arise in the diaphy- seal marrow space, similar in pattern to hematogenous metastatic disease to the bone (11). Contiguous spread of infection from adjacent soft tissues is more prevalent in the adults than in children, although hematogenous spread is still more common (12). Contiguous infections can occur in trauma patients with open fractures, in bedridden patients with decubitus ulcers, and in patients with a diabetic foot. Localizing symptoms are more prevalent in the adult population as opposed to the pediatric population, allowing for more dedicated anatomic imaging with MRI, rather than a survey with radionuclide bone scanning.
Epidemiology
The annual incidence of osteomyelitis in children under 13 years of age is 1/5000 (13). With boys slightly more often affected than girls, fast-growing long bones such as the tibia and femur are the most affected regions.
symptoms and the inherent high spatial resolution allows exact
anatomic detail that may be helpful for surgical planning (limited to
moderate evidence).
Approximately 25% of cases affect the flat bones including the pelvis.
Although a single bone is usually affected, polyostotic involvement has been reported in up to 6.8% of cases in infants and in 22% of neonates (4,14,15). The most common organisms are S. aureus, followed by b- hemolytic Streptococcus, Streptococcus pneumoniae, Escherichia coli, and Pseudomonas aeruginosa (3,16). Clinical presentation can be confusing, and many laboratory findings such as elevated sedimentation rate may be sen- sitive but not specific. Serial blood cultures are only positive in 32% to 60%
of cases (1,17,18). Infections in infants and neonates are usually clinically silent, and toddlers may present with limping, pseudoparalysis, or pain on passive movement (19).
Half of the cases of septic arthritis occur in children less than 3 years of age (20). Approximately 53% are isolated cases of septic arthritis and 47%
are cases of septic arthritis associated with osteomyelitis (21). Conversely, 30% of patients with osteomyelitis have adjacent septic arthritis (22). Boys are slightly more affected than girls (1.2 : 1), and the hip is the most affected joint (23). The most common symptoms are pain, fever, refusal to bear weight, and joint swelling. Most cases involve a single joint, although up to 15% of cases can affect multiple joints. Mortality rates of up to 7% have been reported (21). Similar organisms to those in osteomyelitis are found in septic arthritis, including S. aureus and S. pneumoniae (21,24). The most common sequelae of septic arthritis include joint instability, joint function limitation, and limb shortening (25).
Overall Cost to Society
No data were found in the medical literature on the overall cost to society from the diagnosis, treatment, and complications of acute hematogenous osteomyelitis or septic arthritis. Although there are several cost-effectiveness analyses evaluating the type, extent, and route of antibi- otic administration in the treatment of osteomyelitis and septic arthritis, no cost-effectiveness data were found in the literature specifically incorporat- ing imaging strategies in the management of acute hematogenous osteomyelitis or septic arthritis.
Goals
In acute hematogenous osteomyelitis and septic arthritis, the goal is early diagnosis and treatment to prevent the long-term sequelae of these diseases, which include growth disturbances, joint instability, chronic infection, malalignment, and limb deformity. The standard treatments include intravenous antibiotics and/or surgical debridement. Septic arthri- tis usually requires surgical therapy in order to decompress the intraartic- ular pressure. Surgical debridement may be necessary for osteomyelitis if frank pus can be aspirated from the bone, if there is necrotic bone present, or if there is failure to respond to antibiotic therapy (15,26).
Methodology
The authors performed a Medline search using PubMed (National Library
of Medicine, Bethesda, Maryland) for data relevant to the diagnostic
performance and accuracy of both clinical and radiographic examination
of patients with acute hematogenous osteomyelitis and septic arthritis.
The diagnostic performance of the clinical examination (history and physical exam) and surgical outcome was based on a systematic litera- ture review performed for the years 1966 to 2004. The clinical examina- tion search strategy used the following terms: (1) acute hematogenous osteomyelitis, (2) septic arthritis, (3) pediatric, (4) children, (5) clinical examina- tion, (6) epidemiology or physical examination or surgery, and (7) treatment or surgery. The review of the diagnostic imaging literature was done for the same years. The search strategy used the following key words: (1) acute hematogenous osteomyelitis, (2) septic arthritis, (3) magnetic resonance imaging or MRI, (4) bone scan, (5) ultrasound, and (6) imaging, as well as combina- tions of these search strings. We excluded animal studies and non–English- language articles.
I. What are the Clinical Findings that Raise the Suspicion for Acute Hematogenous Osteomyelitis and Septic
Arthritis to Direct Further Imaging?
Summary of Evidence: The clinical presentation of acute hematogenous osteomyelitis and septic arthritis can be confusing and nonspecific in the pediatric population. No single clinical finding in isolation leads to the diagnosis of osteomyelitis or septic arthritis. Repeat high-resolution imaging may be required to determine the need for surgical debridement, including extension into soft tissues or complications that are not amenable to systemic antibiotic therapy (limited evidence).
Supporting Evidence: Standard laboratory tests such as elevated sedimen- tation rate can be nonspecific or even normal (19) (limited evidence). Serial blood cultures are reported to be positive in 32% to 60% of cases (1,17,18) (moderate and limited evidence). Occasionally, direct aspiration of bone material may be needed for diagnosis. These aspirations can yield positive cultures in 87% of cases (27) (limited evidence).
The clinical presentation in the pediatric age group can be nonspecific.
Infection in the neonate and infant is usually clinically silent. Toddlers can present with limping, pseudoparalysis, or pain on passive movement (4,28) (moderate to limited evidence).
Due to similarities in pathogenesis, there is also overlap in the clinical
presentation of septic arthritis and osteomyelitis. Irritability, limping, or
refusal to bear weight, along with elevated sedimentation rate or leukocy-
tosis, are the most common presentations (15,23,24,29,30) (moderate to
limited evidence). Kocher et al. (30) proposed probabilities for the presence
of septic arthritis in the hip in order to guide further imaging and joint
aspiration based on four clinical variables. These four predictors were
leukocytosis greater than 12,000/mL, fever, inability to bear weight, and
erythrocyte sedimentation rate (ESR) >40mm/hr. If none of these predic-
tors were present, there was a 0.2% chance of septic arthritis. The predicted
probability of septic arthritis with one predictor was 3%, 40% with two pre-
dictors, 93.1% with three predictors, and 99.6% with four predictors. This
constellation of clinical findings was most suggestive of osteomyelitis or
septic arthritis and warranted further evaluation with imaging (moderate
to limited evidence).
II. What Is the Diagnostic Performance of the Different Imaging Studies in Acute Hematogenous Osteomyelitis and Septic Arthritis?
Summary of Evidence: Although plain radiographs are neither sensitive nor specific, their low cost, ready availability, and ability to exclude other diseases that can produce similar symptoms (fractures, tumors) argue for their continued use as the initial evaluation (moderate to limited evidence) (31–35).
Several studies have shown that MRI and radionuclide bone scintigra- phy have high sensitivity for detection of osteomyelitis (moderate evi- dence). Their relative merits have not been established. Bone scintigraphy has the advantage of whole-body imaging when symptoms cannot be localized, but has decreased specificity. This is especially true in the pres- ence of superimposed disease processes such as a joint under pressure, or underlying bone diseases such as sickle cell or Gaucher’s disease (moder- ate to limited evidence) (36–43).
Magnetic resonance imaging has the advantage of higher specificity and higher resolution to evaluate for soft tissue extension or complications, but has limited coverage of the body. This can be a disadvantage if symptoms cannot be localized or if there is polyostotic involvement (moderate to limited evidence).
Ultrasound is highly sensitive for the detection of a joint effusion, but not specific for the presence of infection. Based on the clinical predictors proposed by Kocher et al. (30), a decision to aspirate an effusion can be reliably made to exclude septic arthritis (moderate evidence) (44).
Supporting Evidence: Initial radiographs can detect deep soft tissue swelling and loss of soft tissue planes as early as 48 hours after onset of symptoms, but bone destruction is usually not detectable until 7 to 10 days after onset of symptoms (45). At least 30% of bone destruction is required before osteomyelitis becomes radiographically apparent (2). The sen- sitivity and specificity of plain radiographs are 43% to 75% and 75% to 83%, respectively (limited evidence) (32,46,47). If bone destruction is detected, however, no further imaging may be necessary. In addition, radi- ographs can detect other pathologies such as fractures and tumors that can clinically mimic osteomyelitis (moderate to limited evidence) (31–35,48).
The overall sensitivity and specificity for radionuclide bone scanning are 73% to 100% and 73% to 79% (moderate evidence) (36,41,49–53). In the neonate, however, the sensitivity of radionuclide bone scanning is decreased, ranging from 53% to 87% (54,55). Advantages of bone scin- tigraphy include the ability to image the entire body, delayed imaging with a single administration of tracer, and less sedation requirements.
The ability to image the entire skeleton is ideal if symptoms cannot be localized or if there is polyostotic disease (limited to weak evidence) (33,51,52,56).
The sensitivity and specificity for MRI are 82% to 100% and 75% to 96%
(moderate evidence) (33,57–64). Magnetic resonance imaging has the
advantage of both high sensitivity and specificity. It can also display high-
resolution images and evaluate for complications such as abscesses, joint
effusions, and soft tissue extension that would require surgical interven-
tion (63,65,66). The disadvantages include slighter higher cost relative to bone scintigraphy; prolonged imaging times, which may require sedation;
and limited coverage.
Ultrasound is highly sensitive for the evaluation of joint effusions and can detect as little as 5 to 10 cc of fluid within a joint (67). However, no ultrasound characteristics, including complexity of the fluid, the quantity of fluid, or adjacent hyperemia on color Doppler imaging, have been shown to be definitive in distinguishing septic arthritis from other non- infectious causes of joint effusions (68–71). Despite this limitation, the absence of fluid by ultrasound can be very helpful as septic arthritis is very unlikely in this setting (33,71,72). As outlined above, Kocher et al. (30) have provided clinical guidelines to direct joint aspiration. These include fever, the presence of elevated white count, an elevated sedimentation rate, and inability to bear weight (moderate evidence).
III. What Is the Natural History of Osteomyelitis and Septic Arthritis, and What Are the Roles of Medical Therapy Versus Surgical Treatment?
Summary of Evidence: Most uncomplicated cases of osteomyelitis require hospitalization and the institution of systemic intravenous antibiotic therapy. If there is a delay of more than 4 days prior to institution of therapy, there is increased poor outcomes and long-term sequelae (mod- erate evidence). Approximately 5% to 10% of cases require surgical inter- vention after initial antibiotic therapy, and up to 20% to 50% of all cases eventually require some form of surgery, including reconstruction and repeat debridements.
Approximately 5% to 10% of all cases have long-term sequelae such as growth disturbance, loss of function, malalignment, and deformity.
Approximately 6% of cases develop chronic osteomyelitis (73).
Supporting Evidence: Most cases of acute osteomyelitis and septic arthritis are treated with antibiotics. If frank pus is aspirated from a joint, surgical debridement is required immediately. Patients are admitted for initiation of systemic antibiotic therapy. Average length of stay ranges from 3 to 7 days (16,24,74). Average course of systemic antibiotic therapy is approxi- mately 11 to 14 days with an additional 4 weeks of outpatient oral antibi- otic therapy (5,7,16,75). Many of the clinical signs and symptoms improve within 48 hours of initiation of systemic antibiotics, which is a reassuring sign. If there is no clinical improvement, further evaluation including imaging may be required to exclude complications not amenable to antibi- otics alone, such as abscess collections, necrotic tissue, or extension into soft tissues.
Approximately 20% to 50% of all cases eventually require surgical inter-
vention (28). Up to 10% of patients eventually have long-term sequelae,
including growth disturbance, loss of function, malalignment, and defor-
mity (8,9,16,23,28). Up to 6% of patients eventually have chronic oste-
omyelitis. There is evidence that a delay in initiation of therapy ( >4 days
after onset of symptoms), certain infecting organisms (methicillin-resistant
S. aureus), and age of the patient ( <6 months of age) are predictors of bad
outcomes (moderate evidence) (3,7,16,73).
IV. Is There a Role for Repeat Imaging in the Management?
Summary of Evidence: Most patients respond clinically to systemic anti- biotics within 48 hours. If there is no clinical response to therapy, repeat imaging should be performed to exclude complications that would require surgical intervention such as abscess collections, extensive soft tissue exten- sion, or necrotic tissue. The performance characteristics of MRI are ideal in this setting (moderate to limited evidence).
Supporting Evidence: Approximately 95% to 98% of patients respond clini- cally to antibiotic therapy alone (76). Children usually respond quickly to antibiotics, on average within 48 hours. However, approximately 5% to 10%
of patients eventually require surgical intervention (77,78). These patients require high-resolution imaging to evaluate for surgical disease. The liter- ature supports the use of MRI for evaluation of necrosis, abscess collections, and soft tissue extension (63–65,79) (moderate evidence to limited evi- dence). This information can be helpful for the surgeon in planning the sur- gical approach and method of debridement. There are also some data in the literature suggesting that MRI should be the repeat imaging modality of choice if the site of infection is localized to the spine or pelvis. There is a higher incidence of abscess formation in these deep infections, which would require earlier surgical evaluation and treatment (33,57,63,80).
V. What Is the Diagnostic Performance of Imaging of Osteomyelitis and Septic Arthritis in the Adult?
Summary of Evidence: Overall, MRI appears to be the imaging modality of choice to evaluate for osteomyelitis and septic arthritis in the adult popu- lation, including the diabetic patient and intravenous drug users. The ability to localize symptoms and inherent high spatial resolution allows exact anatomic detail that may be helpful for surgical planning (limited to moderate evidence).
Supporting Evidence: Osteomyelitis in the diabetic foot represents a diag- nostic challenge both clinically and by imaging. The diabetic foot is prone to infection and suboptimal healing due to the decreased blood supply from diabetic vasculopathy, decreased immune response, and repetitive trauma and abnormal mechanics from diabetic neuropathy (81). Because of these abnormalities, there are baseline abnormal imaging findings of the bones and joints without superimposed infection.
Radiographically, the diabetic foot has many features mimicking infec- tion, including destruction, debris, and subluxation. The diabetic foot can also have abnormal findings without osteomyelitis on three-phase radionuclide bone scan (82). There is some evidence of using both bone scan with methylene diphosphanate (MDP) as well as a white blood cell scan to map out specific areas of infection (82–85) (limited to moderate evi- dence). Although it has excellent sensitivity and specificity (92 and 97%
respectively), the technique is cumbersome and laborious (83). Its lower resolution relative to MRI also limits the imaging of anatomic detail for surgical planning (86,87).
Magnetic resonance imaging has both high sensitivity and specificity for
evaluating osteomyelitis in the diabetic foot (88–91). Sensitivity ranges
from 88% to 92% and specificity ranges from 82% to 100% (85,90,92) (mod- erate evidence). However, the diagnosis is frequently not made based on specific imaging characteristics, but by the location of the abnormality.
The neuropathic foot inherently contains signal abnormalities similar to osteomyelitis. Imaging diagnosis is made by identifying signal abnormal- ities in the bone contiguous and in direct contact with adjacent skin ulcers and known pressure points in the diabetic foot (87).
Hematogenous osteomyelitis and septic arthritis also occurs in intra- venous drug users. Many of these infections arise initially in the soft tissues, such as the psoas muscle, with subsequent involvement into the spine or sacroiliac (SI) joint (93,94). Septic arthritis with osteomyelitis is slightly more common in this population than osteomyelitis alone (95). The plain film is neither sensitive nor specific in commonly involved locations, such as the spine and SI joint. Computed tomography (CT) scan with intra- venous contrast material has been shown to be very accurate in the iden- tification of the soft tissue infections and abscesses, but not as accurate in the evaluation of the spinal osteomyelitis/discitis or sacroiliitis (96,97) (limited to moderate evidence). Magnetic resonance imaging is superior in evaluating these structures due to its higher contrast, signal-to-noise ratio, and multiplanar imaging.
Ultrasound can still detect joint effusions, but can be technically more difficult due to the larger amount of soft tissue in adults compared to the pediatric population (98,99). Magnetic resonance imaging is highly sensi- tive for the evaluation of septic arthritis (97,100,101). Hyperemia and syn- ovitis can also be elucidated with the use of intravenous gadolinium, increasing the accuracy of septic arthritis (102).
VI. What Are the Roles of the Difference Imaging Modalities in the Evaluation of Acute Osteomyelitis and Septic Arthritis?
The decision tree in Figure 14.1 outlines the role of each imaging modality in the evaluation of suspected osteomyelitis. Table 14.1 summarizes the diagnostic performance of the imaging studies for osteomyelitis in children and adults. The plain radiograph is the initial imaging evaluation due to its relative low cost, rapid acquisition, and ready availability. If there is frank evidence of osteomyelitis on the radiograph, immediate antibiotic therapy can be instituted and further imaging may not be necessary, as up to 80% of patients are successfully treated with antibiotics alone.
If the radiograph is negative for osteomyelitis, and there are no localiz- ing symptoms clinically, radionuclide bone scintigraphy is the next imag- ing modality, based on its ability to provide whole-body imaging.
If there are localized symptoms, MRI would be a better choice due to higher resolution, more specificity, and ability to immediately evaluate complications.
Repeat imaging with MRI should be considered in all patients who do not improve clinically after 48 hours of systemic antibiotic therapy, and to direct management of those who require surgical therapy. In addition, if immediate surgical therapy is planned, such as in cases of infections involving the spine or pelvis, earlier imaging with MRI may be of use.
If symptoms are referable to the hip, an ultrasound should be performed
to rapidly evaluate for the presence of an effusion, and also to provide
imaging-guided joint aspiration.
Table 14.1 presents the performance characteristics of imaging studies for osteomyelitis in children and adults.
Imaging Case Studies
Case 1
Young child with fever and limp (Fig 14.2).
Case 2
Child with fever (Fig 14.3).
Case 3
Teenager with right buttock pain and fever (Fig 14.4).
Figure 14.1. Algorithm for imaging suspected osteomyelitis or septic arthritis in the pediatric population.
Table 14.1. Diagnostic performance characteristics of imaging studies for osteomyelitis in children and adults
Sensitivity Specificity
Plain radiograph 43–75% 75–83%
(32,46,47)
Radionuclide scintigraphy 73–100% (53–87% in infants) 73–79%
(36,41,49–53)
MRI (33,57–64) 82–100% 75–96%
Figure 14.2. Ultrasound depicting hip effusion with synovitis. Frank pus was aspi- rated from the joint.
Figure 14.3. Radionuclide bone scan shows abnormal uptake in the proximal left tibial metaphysis that was found to be osteomyelitis. The imaging findings are not specific for osteomyelitis, because neoplasms and trauma could have a similar appearance.
Suggested Imaging Protocols
• Plain radiograph: At least two orthogonal views of the body part of
interest should be obtained; views of the opposite limb may be useful
for comparison to detect subtle changes. Imaging should be performed
on all patients suspected of osteomyelitis or septic arthritis to evaluate
for destruction, as well as to exclude other pathologies such as tumors
or fractures.
• Radionuclide bone scintigraphy: Three-phase radionuclide bone scintig- raphy with technetium 99m (Tc-99m)-labeled MDP should be obtained, with planar images during blood flow and soft tissue phases. Planar images of extremities and single photon emission computed tomogra- phy (SPECT) images of the axial skeleton during the bone phase should be obtained. This imaging should be used if symptoms are nonlocaliz- ing or if there is a suspicion of polyostotic disease.
• Magnetic resonance imaging: Axial and coronal T1 spin echo, axial and sagittal T2 fast spin echo with fat saturation, coronal short-time inver- sion recovery (STIR), axial and coronal T1 two-dimensional (2D) spoiled gradient recalled (SPGR) with fat saturation before and after intravenous gadolinium should be obtained. Imaging should be performed if there are localizing symptoms or if the patient fails to respond to antibiotics within 48 hours. Early evaluation with MRI also may be of use if imme- diate surgery is planned.
• Ultrasound: Linear transducer high-frequency probe (7–12 MHz) imag- ing should be obtained and compared with that for the opposite joint to assess symmetry. Color or power Doppler assesses for hyperemia.
Imaging should be performed to evaluate for joint effusion and joint aspiration. It is most commonly used for the hip joint.
Future Research
• Can the use of emerging whole-body imaging techniques in MRI obviate the need for radionuclide scintigraphy in the evaluation of osteomyelitis?
• Can MRI with gadolinium provide more information than ultrasound in the evaluation of septic arthritis?
• Can findings on imaging (plain film, MRI, ultrasound) predict the like- lihood of success of medical therapy alone, and provide early triage to surgical therapy?
• Does positron emission tomography (PET)- CT have a role in the eval- uation of osteomyelitis or septic arthritis?
Figure 14.4. Axial MRI of the pelvis after administration of gadolinium shows abnormal enhancement of the right ischial tuberosity and surrounding soft tissues consistent with osteomyelitis. There is also enhancement around the left greater trochanter, consistent with trochanteric bursitis.
References
1. Waldvogel FA, Medoff G, Swartz MN. N Engl J Med 1970;282(4):198–206.
2. Faden H, Grossi M. Am J Dis Child 1991;145(1):65–69.
3. Kao HC, et al. J Microbiol Immunol Infect 2003;36(4):260–265.
4. Oudjhane K, Azouz EM. Radiol Clin North Am 2001;39(2):251–266.
5. Barton LL, Dunkle LM, Habib FH. Am J Dis Child 1987;141(8):898–900.
6. Azouz EM, Greenspan A, Marton D. Skeletal Radiol 1993;22(1):17–23.
7. Welkon CJ, et al. Pediatr Infect Dis 1986;5(6):669–676.
8. Choi IH, et al. J Bone Joint Surg Am 1990;72(8):1150–1165.
9. Betz RR, et al. J Pediatr Orthop 1990;10(3):365–372.
10. Alderson M, et al. J Bone Joint Surg Br 1986;68(2):268–274.
11. Tice AD, Hoaglund PA, Shoultz DA. J Antimicrob Chemother 2003;51(5):
1261–1268.
12. Jensen AG, et al. J Infect 1997;34(2):113–118.
13. Sonnen GM, Henry NK. Pediatr Clin North Am 1996;43(4):933–947.
14. Asmar BI. Infect Dis Clin North Am 1992;6(1):117–132.
15. Nelson JD. Infect Dis Clin North Am 1990;4(3):513–522.
16. Karwowska A, Davies HD, Jadavji T. Pediatr Infect Dis J 1998;17(11):
1021–1026.
17. Dormans JP, Drummond DS. J Am Acad Orthop Surg 1994;2(6):333–341.
18. Nixon GW. AJR 1978;130(1):123–129.
19. Restrepo SC, Gimenez CR, McCarthy K. Rheum Dis Clin North Am 2003;29(1):
89–109.
20. Matan AJ, Smith JT. Orthopedics 1997;20(7):630–635; quiz 636–637.
21. Caksen H, et al. Pediatr Int 2000;42(5):534–540.
22. Perlman MH, et al. J Pediatr Orthop 2000;20(1):40–43.
23. Wang CL, et al. J Microbiol Immunol Infect 2003;36(1):41–46.
24. Razak M, Nasiruddin J. Med J Malaysia 1998;53(suppl A):86–94.
25. Campagnaro JG, et al. Chir Organi Mov 1992;77(3):233–245.
26. Dagan R. Pediatr Infect Dis J 1993;12(1):88–92.
27. Howard CB, et al. J Bone Joint Surg Br 1994;76(2):311–314.
28. Razak M, Ismail MM, Omar A. Med J Malaysia 1998;53(suppl A):83–85.
29. Khachatourians AG, et al. Clin Orthop 2003;409:186–194.
30. Kocher MS, Zurakowski D, Kasser JR. J Bone Joint Surg Am 1999;81(12):
1662–1670.
31. Zucker MI, Yao L. West J Med 1992;156(3):297–298.
32. Bonakdar-pour A, Gaines VD. Orthop Clin North Am 1983;14(1):21–37.
33. Jaramillo D, et al. AJR 1995;165(2):399–403.
34. Gold R. Pediatr Infect Dis J 1995;14(6):555.
35. Fordham L, Auringer ST, Frush DP. J Pediatr 1998;132(5):906–908.
36. Sullivan DC, et al. Radiology 1980;135(3):731–736.
37. Wald ER, Mirro R, Gartner JC. Clin Pediatr (Phila) 1980;19(9):597–601.
38. Jones DC, Cady RB. J Bone Joint Surg Br 1981;63–B(3):376–378.
39. Berkowitz ID, Wenzel W. Am J Dis Child 1980;134(9):828–830.
40. Handmaker H. Radiology 1980;135(3):787–789.
41. Barron BJ, Dhekne RD. Clin Nucl Med 1984;9(7):392–393.
42. Park HM, Rothschild PA, Kernek CB. AJR 1985;145(5):1079–1084.
43. Stark JE, et al. Radiology 1991;179(3):731–733.
44. Klein DM, et al. Clin Orthop 1997;338:153–159.
45. Capitanio MA, Kirkpatrick JA. AJR Radium Ther Nucl Med, 1970;108(3):
488–496.
46. Kaye JJ. Pediatr Ann 1976;5(1):11–31.
47. Keenan AM, Tindel NL, Alavi A. Arch Intern Med 1989;149(10):2262–2266.
48. Gold R. Pediatr Rev 1991;12(10):292–297.
49. Duszynski DO, et al. Radiology 1975;117(2):337–340.
50. Gelfand MJ, Silberstein EB. JAMA 1977;237(3):245–247.
51. Hankins JH, Flowers WM Jr. J Miss State Med Assoc 1978;19(1):10–12.
52. Nelson HT, Taylor A. Eur J Nucl Med 1980;5(3):267–269.
53. Erasmie U, Hirsch G. Z Kinderchir 1981;32(4):360–366.
54. Ash JM, Gilday DL. J Nucl Med 1980;21(5):417–420.
55. Bressler EL, Conway JJ, Weiss SC. Radiology 1984;152(3):685–688.
56. Handmaker H, Leonards R. Semin Nucl Med 1976;6(1):95–105.
57. Modic MT, et al. Radiol Clin North Am 1986;24(2):247–258.
58. Unger E, et al. AJR 1988;150(3):605–610.
59. Morrison WB, et al. Radiology 1993;189(1):251–257.
60. Fletcher BD, Scoles PV, Nelson AD. Radiology 1984;150(1):57–60.
61. Berquist TH, et al. Magn Reson Imaging 1985;3(3):219–230.
62. Dangman BC, et al. Radiology 1992;182(3):743–747.
63. Mazur JM, et al. J Pediatr Orthop 1995;15(2):144–147.
64. Umans H, Haramati N, Flusser G. Magn Reson Imaging 2000;18(3):255–262.
65. Connolly LP, et al. J Nucl Med 2002;43(10):1310–1316.
66. Lee SK, et al. Radiology 1999;211(2):459–465.
67. Moss SG, et al. Radiology 1998;208(1):43–48.
68. Chao HC, et al. Acta Paediatr Taiwan 1999;40(4):268–270.
69. Chao HC, et al. J Ultrasound Med 1999;18(11):729–734; quiz 735–736.
70. Strouse PJ, DiPietro MA, Adler RS. Radiology 1998;206(3):731–735.
71. Zawin JK, et al. Radiology 1993;187(2):459–463.
72. Lim-Dunham JE, Ben-Ami TE, Yousefzadeh DK. Pediatr Radiol 1995;25(7):
556–559.
73. Espersen F, et al. Rev Infect Dis 1991;13(3):347–358.
74. Speiser JC, et al. Semin Arthritis Rheum 1985;15(2):132–138.
75. Scott RJ, et al. J Pediatr Orthop 1990;10(5):649–652.
76. Le Saux N, et al. BMC Infect Dis 2002;2(1):16.
77. Roine I, et al. Clin Infect Dis 1997;24(5):849–853.
78. Dirschl DR. Orthop Rev 1994;23(4):305–312.
79. McAndrew PT, Clark C. BMJ 1998;316(7125):147.
80. Middleton MS. AJR 1988;151(3):612–613.
81. Snyder RJ, et al. Ostomy Wound Manage 2001;47(1):18–22, 25–30; quiz 31–32.
82. Unal SN, et al. Clin Nucl Med 2001;26(12):1016–1021.
83. Poirier JY, et al. Diabetes Metab 2002;28(6 Pt 1):485–490.
84. Crerand S, et al. J Bone Joint Surg Br 1996;78(1):51–55.
85. Cook TA, et al. Br J Surg 1996;83(2):245–248.
86. Sella EJ, Grosser DM. Clin Podiatr Med Surg 2003;20(4):729–740.
87. Schweitzer ME, Morrison WB. Radiol Clin North Am 2004;42(1):61–71, vi.
88. Craig JG, et al. Radiology 1997;203(3):849–855.
89. Marcus CD, et al. Radiographics 1996;16(6):1337–1348.
90. Morrison WB, et al. Radiology 1995;196(2):557–564.
91. Ledermann HP, Morrison WB, Schweitzer ME. Radiology 2002;223(3):747–755.
92. Croll SD, et al. J Vasc Surg 1996;24(2):266–270.
93. Alcantara AL, Tucker RB, McCarroll KA. Infect Dis Clin North Am 2002;
16(3):713–743, ix–x.
94. Kak V, Chandrasekar PH. Infect Dis Clin North Am 2002;16(3):681–695.
95. Chandrasekar PH, Narula AP. Rev Infect Dis 1986;8(6):904–911.
96. Bonham P. J Wound Ostomy Continence Nurs 2001;28(2):73–88.
97. Sturzenbecher A, et al. Skeletal Radiol 2000;29(8):439–446.
98. Wingstrand H, Egund N, Forsberg L. J Bone Joint Surg Br 1987;69(2):254–256.
99. Zieger MM, Dorr U, Schulz RD. Skeletal Radiol 1987;16(8):607–611.
100. Karchevsky M, et al. AJR 2004;182(1):119–122.
101. Learch TJ, Farooki S. Clin Imaging 2000;24(4):236–242.
102. Graif M, et al. Skeletal Radiol 1999;28(11):616–620.
