Clinical management of shoulder sepsis

Presentation

Pain, loss of motion, and effusion are early signs of infection. However, shoulder effusion is difficult to detect and is often missed. Motion is painful, and the arm is adducted and internally rotated. Systemic signs include fever, leukocytosis, and sedimentation rate changes. The symptoms of immunosuppressed and rheumatoid patients may be muted. Delay in presentation is common, averaging greater than 7 days. In a 50-patient case series by Abdel and colleagues, eight showed bilateral infection.

The sternoclavicular joint may be involved and should be suspected in unilateral enlargement without trauma in patients younger than 50 years. Gonococcal and staphylococcal infections have been reported. Intravenous (IV) drug users are susceptible to infection by gram-negative organisms, especially Pseudomonas aeruginosa and Serratia marcescens . The acromioclavicular joint is rarely involved in sepsis but may be contaminated by steroid injections. Infection caused by intra-articular steroid injection can also progress to an intraosseous abscess ( Fig. 48.1 ).

Wound infection is not uncommon after surgical resection of the distal clavicle for osteoarthritis, probably due to the proximity of the joint to the overlying skin and the lack of an intervening muscle layer. Chanet and colleagues reported that a past history of radiation for breast cancer was a risk factor for the development of septic arthritis of the glenohumeral and sternoclavicular joints. The mean time elapsed since radiation was 16 years (range, 3 to 34 years).

Gabriel and colleagues reported on a 5-week-old boy who had a brachial plexus neuropathy and paralysis of the upper extremity secondary to septic arthritis of the glenohumeral joint and osteomyelitis of the proximal part of the humerus. They pointed out that pseudoparalysis of a limb associated with sepsis was a well-documented phenomenon. Similarly, muscular spasm associated with pain caused by infection can lead to apparent weakness. True nerve paralysis associated with osteomyelitis is uncommon, and its documentation by electrodiagnostic studies is rare. Permanent weakness persisted in the patient even though the glenohumeral joint and proximal humerus were surgically drained. The possible causes of the plexus neuropathy associated with infection included ischemic neuropathy from thrombophlebitis of the vasa nervorum, arterial embolism, hyperergic or hypersensitivity reactions, and local compression due to abscess formation. Rankin and Rycken reported bilateral dislocations of the proximal humerus as a result of septic arthritis. Miron and colleagues have also reported transient brachial plexus palsy associated with suppurative arthritis of the shoulder.

Sepsis after total joint replacement of the shoulder is relatively rare because of prophylactic antibiotic use and the excellent blood supply of the surrounding soft tissues. Symptoms of infection include pain, loss of motion, and subluxation. Pain relief is so universal after total shoulder replacement that infection should be suspected if there is pain or lucency around the cement mantle. Although the risk of prosthetic infections is very low in asymptomatic patients, the ubiquitous nature of Cutibacterium acnes has elevated our concern regarding its detection. However, C. acnes has a high positivity rate in asymptomatic patients, and routine sampling is discouraged without clinical evidence for infection.

Patients with chronic rheumatoid arthritis are susceptible to spontaneous septic arthritis. ^, Because the active destructive process of the rheumatoid arthritis masks the septic condition, detection of the infection is often delayed. The onset of septic arthritis should be suspected when the clinical course of the rheumatoid patient worsens acutely, especially if the disease is long term. When infection is present, the patient experiences a sudden aggravation of pain as well as swelling and increased temperature in the joint; sudden chills can also occur. The physician should emphasize to patients with chronic rheumatoid arthritis (and to caregivers) that a sudden exacerbation of symptoms warrants investigation.

Radiographs are not especially useful in the early diagnosis because septic arthritis does not significantly alter the bone destruction resulting from rheumatoid arthritis and because bone and joint radiographic changes are delayed. Other acute arthritic disorders can imitate or mask sepsis, including gout, pseudogout, rheumatic fever, juvenile rheumatoid arthritis, neuropathic arthropathy, and the oligoarthritic syndromes. Trauma and tumors can cause adjacent joint effusions and must be considered.

Evaluation

The culture and analysis of synovial joint fluid is critical for diagnosis. The joint should be aspirated when septic arthritis is suspected. Radiographic guidance is indicated if the approach and access is difficult especially in total joints. When the results of cultures are negative and the diagnosis is difficult to determine, an arthroscopic biopsy can be useful. A synovial biopsy and culture for acid-fast organisms and fungi should be performed in patients with chronic monoarticular arthritis, especially those with tenosynovitis.

Viral infection must be considered when bacteria cannot be identified. Smith and Piercy report that viral arthritis is associated with rubella, parvovirus, mumps, hepatitis B, and lymphocytic choriomeningitis. For a patient who presents with multiple joint involvement and systemic manifestations consistent with a viral infection, serologic confirmation of the infection should be obtained because it is not usually possible to isolate the virus from joint fluid.

Synovial fluid analysis

If septic arthritis is suspected, the synovial fluid should be examined ( Table 48.1 ). Septic arthritis is probable if the leukocyte count is greater than 50,000 cells/mm ³ , the glucose level is low, and more than 75% of the cells are polymorphonuclear. These findings are beyond the range compatible with uncomplicated rheumatoid arthritis. In bacterial infections aspiration can yield 10 mL or more of fluid. Synovial joint fluid is usually opaque or brownish, turbid, and thick, but it may be serosanguineous in 15% of cases with poor mucin clot formation. Proteins are elevated primarily because of an elevated white blood cell (WBC) count (usually >50,000/mm ³ and often as great as 100,000/mm ³ , primarily of neutrophils). Half of adults and less than half of children have a glucose level in the joint fluid that is 40 mg/L less than that in the serum glucose drawn at the same time. These findings are more common later in infection. Polymorphonuclear leukocytes are dominant (90%); counts greater than 100,000/mm ³ are typical of staphylococcal and acute bacterial infection. Mehta and colleagues reported positive cultures in 96% of aspirates from patients with suspected septic arthritis when the polymorphonuclear differential count was greater than 85% of the total aspirate WBC count. Monocytes are more predominant in mycobacterial infections. Crystal examination is needed to rule out gout or pseudogout. Rheumatoid, rheumatic, and crystalline joint diseases also elevate leukocytes, but the presence of these diseases does not exclude concomitant sepsis.

TABLE 48.1

Synovial Fluid Findings in Acute Pyogenic Arthritis

From Schmid FR. Principles of diagnosis and treatment of bone and joint infections. In: McCarty DJ, ed. Arthritis and Allied Conditions: A Textbook of Rheumatology . Philadelphia: Lea & Febiger; 1985:1638.

	INFLAMMATORY FLUIDS
Joint Fluid Examination	Noninflammatory Fluids	Noninfectious	Infectious
Color	Colorless, pale yellow	Yellow to white	Yellow
Turbidity	Clear, slightly turbid	Turbid	Turbid, purulent
Viscosity	Not reduced	Reduced	Reduced
Mucin clot	Tight clot	Friable	Friable
Cell count (per mm ³)	200–1000	3000–10,000	10,000–100,000
Predominant cell type	Mononuclear	PMN	PMN
Synovial fluid to blood glucose ratio	0.8–1.0	0.5–0.8	<0.5
Lactic acid	Same as plasma	Higher than plasma	Often very high
Gram stain for organism	None	None	Positive ^a
Culture	Negative	Negative	Positive ^a

^aIn some cases, especially in gonococcal infection, no organisms may be demonstrated.

PMN , Polymorphonuclear neutrophil.

The results of Gram staining are positive approximately 50% of the time, but false-positive results do occur. Positive joint cultures occur in 90% of patients with established bacterial septic arthritis and in 75% of patients with tubercular arthritis. Blood cultures should also be obtained; the results are positive in approximately 50% of patients with acute infection. Some prosthesis-centered infections are difficult to detect unless tissue is biopsied and prepared for culture. Protocols may need adjustment to allow adequate growth time to identify slower growing organisms, such as C. acnes . The current recommendations for isolating C. acnes state that it requires at least 13 days after collection. Culture specimens should be taken for gram-positive, gram-negative, aerobic, and anaerobic bacteria and for mycobacteria and fungi. Laboratory technique and media selection should be based on the type of antibiotic given to the patient and on the specific nutrient requirements of suspected bacteria. The magnitude of anaerobic septic arthritis has been underestimated in the past.

Newer testing for infection, including serum markers for interleukin-6, alpha defensin, and leukocyte esterase, shows promise, but clinical testing has shown mixed results.

Imaging studies

Radiographs and computed tomography (CT) scans of the septic shoulder can indicate changes ranging from widening to subluxation and from bone destruction to new bone formation ( Figs. 48.2–48.4 ). Ultrasound is useful in assisting aspiration and in assessing the infected shoulder joints. Technetium bone scanning has been reported to be positive in 75% to 100% of septic arthritis cases, but technetium, gallium, and indium scans are not always consistent. Schmidt and colleagues found that the results of technetium bone scans performed on children with septic arthritis were often negative. Indium scans may be more accurate indicators of sepsis, but conclusive data are lacking. Indium scintigraphy for osteomyelitis should be preceded by a positive result on a technetium scan. If the result of an indium scan is negative, infection is unlikely; a positive result on an indium scan increases the specificity of the diagnosis. Indium uptake should be evaluated against the normal reticuloendothelial background. False-negative results can occur in neonates and during the acute phase of osteomyelitis.

Gupta and Prezio found that the specificity of nuclear scintigraphy using ^99m Tc phosphonates, Ga citrate, or ¹¹¹ In-labeled leukocytes for diagnosing musculoskeletal infection could be improved if two nucleotide studies were used in conjunction. The main limitation of scintigraphic investigations is their limited spatial resolution, making the results inexact. Furthermore, such studies can take hours to days for completion.

CT is useful in identifying small early lytic lesions caused by osteomyelitis that might be obscured in ordinary radiographs. The diagnosis of sternoclavicular joint sepsis and clavicular osteomyelitis infections may be improved using CT because it overcomes the tissue-overlap problem that occurs with ordinary radiographs.

Magnetic resonance imaging (MRI) can facilitate the differentiation of acute from chronic osteomyelitis and can help to detect evidence of active infection in the presence of chronic inflammation, rheumatologic disorder, or posttraumatic lesions. ^, Early changes of osteomyelitis cause accumulation of fluid, inflammatory cells, and exudate in the marrow, producing a focus of low signal intensity within bright, fatty marrow. In chronic osteomyelitis large areas of abnormal or low signals on MRI can indicate an area of possible sequestration and hyperemia, especially along sinus tracts. The capsular distention of articular cartilage and fluid-filled spaces are clearly visible on MRI, as are damaged surfaces, loose bodies, and avascular regions. MRI can be sensitive in early detection and specific for localizing and identifying sequestra ( Figs. 48.5 and 48.6 ).

De Boeck and colleagues reported on the usefulness of MRI in detecting pyomyositis and found it to be useful in excluding other pathologic processes, such as infectious arthritis, osteomyelitis, hematoma, thrombophlebitis, and malignant tumor. MRI is helpful for pregnant patients because it provides a highly sensitive method of detecting skeletal infection without exposing the fetus to ionizing radiation.

Ultrasound can also be used to advantage in defining abscess formation. Widman and colleagues reported that ultrasound can be used to differentiate septic arthritis of the acromioclavicular joint from septic arthritis of the shoulder joint.

Hopkins and colleagues assessed the role of gadolinium-enhanced MRI in providing diagnostic information beyond that given by nonenhanced MRI in the evaluation of musculoskeletal infectious processes. They found that gadolinium-enhanced MRI was a highly sensitive technique that was especially useful in distinguishing abscesses from surrounding cellulitis or myositis. Lack of contrast enhancement ruled out infection with a high degree of certainty. However, these authors pointed out that contrast enhancement cannot be used to reliably distinguish infectious from noninfectious inflammatory conditions. In their study, gadolinium-enhanced MRI was found to have a very high sensitivity (89% to 100%) and high accuracy (79% to 88%) in the diagnosis of various infectious lesions in the musculoskeletal system. However, the specificities (46% to 88%) were not as high ( Fig. 48.7 ).

Tehranzadeh and colleagues have reported on the use of MRI in diagnosing osteomyelitis. In several comparative studies MRI has been reported to be more advantageous in detecting the presence and determining the extent of osteomyelitis compared with scintigraphy, CT, and conventional radiography.

Treatment

Most clinicians agree that systemic antibiotic therapy should begin immediately after the diagnosis of septic arthritis, but there is less agreement on subsequent therapy. Repeated needle aspiration has been recommended as a primary treatment. ^, In our opinion, drainage by urgent arthroscopy or arthrotomy is the treatment of choice, especially for the shoulder. ^, Forward and Hunter have reported the use of a small arthroscope in the treatment of septic arthritis of the shoulder in infants. They recommended insertion of the arthroscope posteriorly and reported full recovery with a single intervention in three patients. In a study of 78 joints (including 10 shoulders). Stutz and colleagues reported that septic arthritis could be treated with arthroscopic irrigation and systemic antibiotic therapy in 91% of the affected joints. Jerosch and colleagues reported the successful treatment of septic arthritis by arthroscopy in 12 patients between 4 and 57 years of age. Jeon and colleagues reported on 19 patients who underwent arthroscopic irrigation and debridement for septic arthritis of the glenohumeral joint. Fifteen patients had received an injection into the joint preceding the development of infection. The infection was eradicated completely with a single procedure in 14 patients. Abdel and colleagues found that 32% (16 of 50) patients required reoperation within 1 month due to recurrent infection.

In general, the literature suggests that treatment of septic arthritis with repeated needle aspiration and appropriate IV antibiotics may be adequate, except for the hip. However, conclusive studies of initial surgical drainage versus needle aspiration are lacking. In a retrospective study comparing 55 infected joints treated by needle aspiration with 18 joints treated surgically, it was concluded that 60% of surgically treated patients experienced sequelae, whereas 80% of medically treated patients recovered completely. However, it must be noted that there was no difference in outcome in the only randomized prospective study that compared aspiration with arthrotomy. In this study of 61 children in Malawi who had septic arthritis of the shoulder, both groups received antibiotics for 6 weeks, and most of the infections were due to Salmonella .

Most orthopedic surgeons believe that the anatomy of the shoulder and the nature of shoulder sepsis demand surgical treatment. Septic arthritis that does not rapidly and progressively respond to medical management should be surgically drained. Sternoclavicular joint infections in drug abusers usually involve both bone and joint. For these patients, needle aspiration has not been useful in establishing the diagnosis. However, surgical drainage has been useful in identifying the pathogen. It also allows debridement of necrotic bone and permits drainage of abscesses, which are often present. ^, Intra-articular sepsis should be treated in a timely manner. Variables that influence the selection of treatment methods include the duration of infection, host immune status, types of infecting organisms, and presence of foreign bodies, or adjacent osteomyelitis; of these, the most critical factors are the infecting organism and the presence of a foreign body.

Osteomyelitis of the proximal metaphysis can precede or be secondary to septic arthritis of the glenohumeral joint. Therefore drilling of the metaphysis has been recommended for pediatric septic glenohumeral arthritis to rule out osteomyelitis and to allow adequate decompression of the bone.

Immobilization is usually recommended for septic arthritis. However, a study by Salter and colleagues of septic arthritis in a rabbit model indicated that the use of continuous passive motion was superior to immobilization or to intermittent active motion. Possible explanations for this include prevention of bacterial adhesion, enhanced diffusion of nutrients, improved clearance of lysosomal enzymes and debris from the infected joints, and stimulation of chondrocytes to synthesize the various components of the matrix.

Gonococcal septic arthritis may be polyarticular. Fever is low grade, usually less than 102°F; articular and periarticular structures become swollen, stiff, and painful, followed by desquamation of skin over the joints. During septicemia, a macular rash or occasionally a vesicular rash occurs in one-third of patients, but such a rash can also be caused by Neisseria meningitidis , Haemophilus species, and Streptobacillus species. Organisms are not common in joint fluid and therefore are not isolated in most cases. The general picture of gonococcal septic arthritis is less acute than that of staphylococcal infection. Gonococcal arthritis responds well to systemic antibiotic therapy and needle aspiration; however, in refractory cases, arthrocentesis may be required.

Antimicrobial agents used to treat joint infections generally achieve intra-articular levels that are equal to or greater than serum levels, except for erythromycin and gentamicin ( Table 48.2 ). Thus intra-articular antibiotics are generally not advocated. In osteomyelitis antibiotic penetration into bone is unreliable, with clindamycin achieving higher bone levels than cephalothin or methicillin. When treating healthy (nonsequestrated) bone in children, delivery of antibiotics to the infected site is likely. However, the opposite situation exists in adults, older people, vascularly compromised patients, and patients with chronic osteomyelitis, sequestration, and sinus formation. Animal studies have suggested that antibiotic combinations (e.g., oxacillin and aminoglycosides) may be more effective than single antibiotics in osteomyelitis. The quinolone group of antibiotics may be useful because of its broad spectrum of activity against staphylococci, Pseudomonas species, and gram-negative bacteria. These agents achieve excellent local levels in living and dead bone. Mutant strains of organisms can emerge, but they can be reduced by the intelligent use of quinolones in combination with other antibiotics.

TABLE 48.2

Empiric Antibiotic Therapy for Septic Arthritis

Modified from American Medical Association, Division of Drugs and Toxicology: Anti-microbial therapy for common infectious diseases. In American Medical Association: Drug Evaluations Annual 1995. Chicago: American Medical Association; 1995:1277–1282.

	ANTIBIOTIC
Pathology	Preferred	Alternative
<2 Months
No organism	Oxacillin and aminoglycoside	Cefazolin and aminoglycoside
Gram-positive cocci	Oxacillin or nafcillin	Cephalothin or cephapirin
Gram-negative bacilli	Aminoglycoside	Cefotaxime
2 Months to 5 Years
No organism	Nafcillin and chloramphenicol	Cefuroxime
Gram-positive cocci	Nafcillin or oxacillin	Cefazolin, cephalothin, vancomycin, clindamycin
Gram-negative bacilli	Aminoglycoside	Cefotaxime
Gram-negative coccobacilli	Cefotaxime or ampicillin and chloramphenicol	Chloramphenicol alone; trimethoprim-sulfamethoxazole
5–15 Years
No organism	Nafcillin or oxacillin	Cefazolin or vancomycin
Gram-positive cocci	Nafcillin or oxacillin	Cefazolin, cephalothin, vancomycin, clindamycin
Gram-negative bacilli ^a	Aminoglycoside	Cefotaxime
Gram-negative cocci	Treat for Neisseria gonorrhoeae
>40 Years
No organism	Nafcillin ± aminoglycoside	Cefazolin ± aminoglycoside
Gram-positive cocci	Nafcillin or oxacillin	Cefazolin, cephalothin, vancomycin, clindamycin
Gram-negative bacilli	Ticarcillin + aminoglycoside	Cefotaxime ± aminoglycoside