Use of Imaging
Limited evidence supports the use of medical imagining in the diagnostic evaluation of patients with a suspected organ/space (i.e. bone, joint, and implant) surgical site infection.

Rationale

Radiography
There was one moderate quality study (Bernard 2004) evaluating the use of radiography for patients with suspected hip and knee prosthesis infection. This study showed poor agreement of radiography with the confirmed infections as both a “rule in” test and a “rule out” test. However Radiography is widely available and inexpensive relative to other imaging modalities, and the consensus of the workgroup is that radiographs be considered as the initial imaging exam for suspected cases of bone and/or implant infection interpreted in combination by a provider with skill and experience in interpretation of musculoskeletal radiographs to assess any and all radiographic features of infection, or other causes of the patient’s symptoms, without commenting on or recommending any single finding or combination of findings.

Radiolabeled Leukocyte Imaging
There were five high quality (Scher 2000, Simonsen 2007, Rand 1990, Pelosi 2004, Joseph 2001) and 10 moderate quality (Pons 1999, Glithero 1993, Kim 2014, Love 2004, Segura 2004, Chik 1996, El Espera 2004, Bernard 2004, Fuster 2011, Wolf 2003) studies evaluating the use of radiolabeled leukocyte imaging (with Indium-111 or Tc-99m hexamethylpropyleneamine oxine) for patients with suspected surgical site infections, predominantly patients with suspected hip and knee prosthesis infection. The duration of time between the initial surgery and the performance of the scan was either unclear or greater than 1 year, on average, for most of these studies. These studies showed inconsistent agreement of radiolabeled leukocyte imaging with the confirmed infections as “rule in” and “rule out” tests. For example, two high quality studies (Scher 2000, Simonsen 2007) showed moderate-strong agreement of radiolabeled leukocyte imaging with the reference standard as a “rule in” test, and weak-moderate agreement as a “rule out” test, for patients with suspected infected hip prostheses. By contrast, two high quality studies (Rand 1990, Scher 2000) showed only weak-moderate agreement of radiolabeled leukocyte imaging with the confirmed infections as a “rule in” test for patients with suspected infected knee prostheses. Stronger agreement with the reference test might be possible in combination with single photon emission computed tomography (Kim 2014) or Tc-99m sulfur colloid bone marrow scintigraphy (Love 2004). The consensus of the workgroup is that radiolabeled leukocyte imaging can be useful as a diagnostic tool (i.e., among other tests) as a “rule in” or “rule out” test for prosthetic joint infection, but its routine use for diagnosis of such infection is not justified, as there may be difficulties diagnosing insidious infections or differentiating infection from aseptic loosening. When radiolabeled leukocyte imaging is performed, the addition of bone marrow scintigraphy can help increase specificity.

Tc-99m-Diphosphonate Skeletal Scintigraphy (“Bone Scan”)
There were three moderate quality studies (Nagoya 2008, Battaglia 2011 and Segura 2004) evaluating the use of skeletal scintigraphy for patients with suspected hip and knee prosthesis infection occurring, on average, greater than one year following surgery. Although one study (Segura 2004) showed strong agreement of two-phase (blood pool and delayed imaging) skeletal scintigraphy with the confirmed infections as a “rule out” test, other studies showed moderate (Nagoya 2008, 3-phase scintigraphy) or weak (Battaglia 2011, unknown number of phases) agreement. Two of the studies (Battaglia 2011, Segura 2004) showed poor agreement of skeletal scintigraphy with the confirmed infections as a “rule in” test, while Nagoya 2008 showed moderate agreement of (3-phase scintigraphy) as a “rule in” test. Skeletal scintigraphy can be useful as a diagnostic tool (i.e., among other tests) as a “rule out” test for delayed (>1 year) prosthetic joint infection if radiolabeled leukocyte imaging is not available, but its role in the diagnosis of such infection is limited.

Positron Emission Tomography (PET) Imaging
There were four high quality (Chacko 2002, Chryssikos 2008, Aksoy 2014, DeWinter 2003) and 3 moderate quality (Kobayashi 2011, Love 2004, Wenter 2015, 2017) studies evaluating the use of F-18 fluorodeoxyglucose (FDG) PET imaging for patients with suspected surgical site infections, with most patients having suspected infection of orthopaedic implants including joint prostheses. The duration of time between the initial surgery and the performance of the scan was either unclear or greater than one year, on average, for most of these studies. These studies showed inconsistent agreement of PET imaging with the confirmed infections as a “rule in” test. For example, two high quality studies (Chacko 2002, Chryssikos 2008) showed strong agreement of FDG-PET imaging with the confirmed infections as a “rule in” test for prosthetic hip joint infection when uptake at the prosthesis-bone interface was used as the criterion for infection. However, another high-quality study (Aksoy 2014) showed that 39/39 hip and knee prostheses with aseptic loosening also showed increased FDG uptake, resulting in poor agreement with the confirmed infections as a “rule in” test for infection. A couple of high quality studies (Chacko 2002, DeWinter 2003) suggest that FDG-PET imaging may be useful as a “rule out” test, showing strong agreement with the confirmed infections; however, other studies show inconsistent results. Better agreement with the reference test might be possible in combination with computed tomography (Wenter 2015). The presence of metallic implants may also affect the diagnostic ability of FDG PET; one high quality study (DeWinter 2003) evaluating the use of FDG PET imaging for patients with suspected surgical site infections of the spine showed weak agreement of PET with the confirmed infections as a “rule in” test when metallic implants were present, but strong agreement as a “rule in” test when implants were not present. Although FDG-PET can be useful as a diagnostic tool (i.e., among other tests) as a “rule in” or “rule out” test for infection, its availability, expense, and current issues with reimbursement are limiting factors, and its routine use is not justified in this setting.

Cross-Sectional Imaging (Magnetic Resonance Imaging, Computed Tomography, Ultrasonography)
There is a lack of data regarding the use of cross-sectional imaging for the diagnosis of orthopaedic surgical site infection. Two moderate quality studies (Li 2016, Plodkowski 2013) evaluating the use of magnetic resonance imaging (MRI) for patients with suspected knee prosthesis infection showed moderate-strong agreement of MRI with the confirmed infections as a “rule in” test, but poor-moderate agreement as a “rule out” test. However, artifacts caused by metallic implants can be significant and limit detection of adjacent bone and soft tissue infection on MRI as well as on computed tomography (CT). Cross-sectional imaging can potentially be useful as a diagnostic tool in certain patients with suspected surgical site infection (e.g., to identify soft-tissue fluid collections) or to guide aspiration/biopsy procedures in such patients, but the potential value of a given imaging exam should be considered on a case-by-case basis.