Osteomyelitis occurs as a result of hematogenous dissemination or contiguous spread of bacteria. Hematogenous osteomyelitis in adults most commonly affects vertebral bodies, although involvement of other sites has been described, particularly in persons who inject drugs. Contiguous-spread osteomyelitis may arise from direct contamination (fracture, joint replacement, orthopedic implant), wounds (pressure sores, diabetic foot ulcers), or adjacent soft tissue infection. Population-based studies suggest that the incidence of osteomyelitis among adults is increasing in the United States, most likely because of the increasing prevalence of diabetes mellitus. Osteomyelitis can be difficult to diagnose, can cause indolent infections that persist for prolonged periods, and requires long-term antibiotic treatment; thus, the economic impact of this infection is substantial.
Osteomyelitis frequently presents as subacute or chronic pain over the affected region of bone. If osteomyelitis has resulted from direct contamination of a wound, the wound may fail to heal or may reopen after healing. Spontaneously opening wounds accompanied by drainage (sinus tracts) are a late manifestation of infection. Underlying osteomyelitis should be considered when chronic wounds, such as pressure ulcers, do not respond to appropriate therapy. Fever and other systemic manifestations of infection are not common but are more likely in patients with acute hematogenously disseminated infection. Clinical findings in patients with diabetes-associated foot ulcer osteomyelitis and vertebral osteomyelitis are discussed separately.
Laboratory studies are nondiagnostic for osteomyelitis. Elevated inflammatory markers, such as erythrocyte sedimentation rate or C-reactive protein level, increase the pretest probability of infection and can be useful in monitoring therapeutic response; normal inflammatory markers alone are insufficient to exclude the diagnosis. Except in acute hematogenous osteomyelitis, leukocytosis is uncommon; in chronic osteomyelitis, anemia may be present. Blood culture results are rarely positive, except in patients with hematogenous osteomyelitis (such as vertebral osteomyelitis). Blood cultures should be obtained when hematogenous osteomyelitis is suspected or in patients with systemic manifestations of sepsis.
Plain radiography is not adequately sensitive to exclude a diagnosis of osteomyelitis, but it is recommended during initial evaluation because of the relatively low cost. The specificity of plain radiography is sufficient to confirm the diagnosis in most patients. If a plain radiograph is not diagnostic, MRI, with and without intravenous contrast, is preferred. If MRI cannot be obtained because of specific contraindications, CT with intravenous contrast is an alternative. Nuclear medicine studies are less sensitive and specific for osteomyelitis but can be used when neither CT with contrast nor MRI are possible after review with the consulting radiologist.
Obtaining biopsy material for culture and pathologic examination is essential to the evaluation of suspected osteomyelitis. Confirming the presence of a pathogen maximizes the chance that the chosen antibiotic therapy will be successful. Specimens may be obtained at surgery or by image-guided biopsy. A bone biopsy is generally not required in persons with positive blood culture results. A possible exception is injection drug users because they have frequent bacteremias, and the organism in the blood culture may not represent the pathogen in the bone. In culture-negative disease, additional testing of biopsy material with nucleic acid amplification techniques, such as broad-range 16S ribosomal RNA gene amplification, may yield the causative organism, although these techniques will not provide information regarding antimicrobial susceptibilities.
Antibiotic therapy for osteomyelitis should be based on results of susceptibility testing from bone or blood culture isolates and knowledge of antibiotic levels achievable in bone for the selected agent. Unless systemic signs of sepsis or concomitant soft tissue infection or bacteremia are present, empiric antibiotics should be withheld until a bone biopsy is obtained. Surgical debridement is indicated if bone necrosis is extensive. Orthopedic hardware should be removed, if possible, to increase the chance of therapeutic success. Parenteral antimicrobial agents are usually chosen initially, but highly bioavailable oral agents with good bone penetration, such as fluoroquinolones, may be considered. Rifampin should be used in combination with another antistaphylococcal agent to manage Staphylococcus aureus infections in the setting of orthopedic hardware if the hardware cannot be removed. Little evidence is available to guide recommendations on duration of therapy; 4 to 6 weeks of antibiotics is considered sufficient for acute infections, whereas longer courses may be required for chronic infections. In some circumstances, especially when hardware cannot be removed, indefinite suppressive therapy may be required. Patients should be informed about the risk of relapse and told that relapse can occur many years after therapy completion.
The incidence of diabetic foot infection is increasing with the increasing prevalence of diabetes mellitus both in the United States and worldwide (see MKSAP 18 Dermatology). These infections are the most frequent diabetes-related complication necessitating hospitalization, cause significant morbidity (especially limb amputation), and are associated with increased mortality. Nonhealing ulcers become colonized with bacteria, after which infection may develop with contiguous spread to bone.
A diagnosis of osteomyelitis should be considered when a diabetic foot ulcer is deep (presence of exposed bone), large (>2 cm in diameter), or chronic (nonhealing after 6 weeks of standard care). Up to two thirds of affected patients do not have leukocytosis or elevated inflammatory markers. A probe-to-bone test (sterile probe inserted into the ulcer base to evaluate for contact with a hard or gritty surface representing bone or joint capsule) should be performed. In a clinically infected ulcer (presence of pus), the positive predictive value of the probe-to-bone test is high; in a noninfected ulcer, the negative predictive value is high. Imaging options are as described for other causes of osteomyelitis. All patients with a new diabetic foot infection should have plain radiography to assess for bony abnormalities, soft tissue gas, and foreign bodies.
Bone samples for histologic confirmation of diagnosis and for culture can be obtained during bone debridement; if debridement is not required, a bone biopsy should be obtained. S. aureus and streptococcal species account for most osteomyelitis complicating diabetic foot ulcers; gram-negative organisms are found in as many as 25% of infections. Anaerobes are much less common. Infections may be polymicrobial.
Although deep sinus-tract tissue cultures can be obtained, the correlation with bone biopsy samples is variable, and bone biopsy remains the recommended modality for microbiologic diagnosis. Recent evidence shows that patients treated with antibiotics chosen on the basis of bone culture results have a better outcome than those treated without these results. If infected but viable bone is present, a 4- to 6-week course of parenteral or oral antibiotic therapy is recommended, as it is for other forms of osteomyelitis. Prolonged oral therapy is indicated if there is residual necrotic bone.
Indications for amputation include persistent sepsis, inability to tolerate antibiotic therapy, progressive bone destruction despite appropriate therapy, or bone destruction that compromises the mechanical integrity of the foot. The patient may also choose amputation over prolonged antibiotic therapy. Hyperbaric oxygen therapy, growth factors, and topical negative-pressure therapy have insufficient evidence of benefit to recommend their use. However, limb salvage may be possible far more often than previously thought when treatment is directed by a dedicated multidisciplinary team consisting of a foot surgeon, a vascular surgeon, an internist, an infectious diseases specialist, nurses, and a physical therapist.
Except when resulting from surgical instrumentation, vertebral osteomyelitis is almost exclusively secondary to hematogenous dissemination. Risk factors for vertebral osteomyelitis include older age, immunocompromise, indwelling catheters, hemodialysis, and injection drug use. Infection occurs in the intervertebral disk space and then spreads to the adjacent vertebral bodies (spondylodiskitis). The lumbar spine is most frequently involved, followed by the thoracic and then the cervical spine. Most infections are due to S. aureus, but S. lugdunensis is increasingly implicated. Persistent bacteremia with other coagulase-negative staphylococci in patients treated with hemodialysis and those with intravascular devices may obviate the need for biopsy. Enterobacteriaceae, Pseudomonas aeruginosa (especially in persons who inject drugs), and Candida species also may cause vertebral osteomyelitis.
New-onset back or neck pain or progressive worsening of chronic pain that is unresponsive to conservative management should raise concern for vertebral osteomyelitis, especially when accompanied by elevated levels of inflammatory markers, neurologic findings, or unexplained fever. Neurologic findings can include sensory loss, weakness, or radiculopathy and are reported in up to one third of patients. Point tenderness is present in as few as one third of patients. Delay in diagnosis is common; in a third of patients, pain is initially attributed to degenerative disease.
As with other forms of osteomyelitis, MRI is the preferred imaging modality. Blood cultures should be performed in all patients. Testing for Mycobacterium tuberculosis infection (with tuberculin skin testing or an interferon-γ release assay), fungal blood cultures, and serologic tests for Brucella species are appropriate for patients at risk for these pathogens (see Mycobacterium tuberculosis Infection, Fungal Infections, and Travel Medicine). A positive Brucella serologic result in the correct epidemiologic setting is considered diagnostic, and biopsy is not needed. Otherwise, image-guided biopsy has a diagnostic yield of approximately 60% and should be used in patients with negative blood culture results. A second biopsy should be obtained if the first is not diagnostic. Nucleic acid amplification techniques can increase biopsy yield; specimens should also be sent for pathologic examination. Open biopsy or percutaneous endoscopic diskectomy and drainage may be considered if the microbiologic diagnosis remains elusive after a second image-guided biopsy attempt.
Patients with neurologic compromise or evidence of spinal instability should undergo evaluation for immediate surgical intervention. Patients with complications, such as severe sepsis, progressive neurologic deficit, spinal instability, or epidural abscess, should receive empiric antibiotic therapy. Otherwise, initiation of antibiotic therapy for uncomplicated vertebral osteomyelitis is based on culture results. Parenteral therapy is generally recommended, especially for S. aureus. However, oral agents with high bioavailability and good bone penetration (such as fluoroquinolones) may be used, especially for Enterobacteriaceae. Certain patient-specific conditions are associated with less common bacterial organisms, including Salmonella osteomyelitis in persons with sickle cell disease and _Pseudomonas aeruginosa_ bone infection in injection drug users. Lengthy parenteral antimicrobial therapy is the mainstay of treatment. An antistaphylococcal agent with methicillin-resistant S. aureus (MRSA) coverage (vancomycin) and a β-lactam with antipseudomonal coverage (cefepime) is an appropriate empiric regimen for suspected osteomyelitis in this patient who uses injection drugs.
The duration of antibiotic therapy for vertebral osteomyelitis is typically 6 weeks. Patients should be followed clinically for improvement in symptoms, and inflammatory markers can be monitored. Repeat imaging, especially MRI, should be reserved for patients who do not respond clinically; worsening of imaging findings in patients with a satisfactory clinical response is well described.
There is no evidence that antibiotics beyond 6 weeks will enhance therapeutic success. In a recently published clinical trial comparing 6 weeks to 12 weeks of antibiotics for vertebral osteomyelitis, cure rates were not significantly different between the treatment groups. Several retrospective studies have found similar results.