Brian F. Mandell M.D., Ph.D.1
1Education Program Director, Department of Rheumatic and Immunologic Diseases, and Senior Associate Program Director, Internal Medicine Residency Program, Cleveland Clinic Foundation
The author has no commercial relationships with manufacturers of products or providers of services discussed in this chapter.
Bacterial infections account for less than 20% of all cases of acute monoarticular and oligoarticular arthritis. Crystal-induced arthritis is approximately four times more common. Because septic arthritis represents a potential threat to life and limb, the possibility of infection dictates the sequence and pace of the diagnostic evaluation. This chapter reviews the clinical presentation, diagnosis, microbiology, and treatment of joint infections. Lyme disease, which can manifest as monoarticular or oligoarticular arthritis of the large joints, is discussed elsewhere [see 7:VII Leptospirosis, Relapsing Fever, Rat-Bite Fever, and Lyme Disease].
Case series of patients with septic arthritis include increasing numbers of immunosuppressed and elderly patients with significant comorbidity [see Clinical Subgroups at Risk for Septic Arthritis, below]. Large retrospective studies from Europe indicate a consistent male-to-female ratio of approximately 1:1, a mean age of approximately 55 years, and polyarticular involvement in up to 15% of cases.1,2,3 These data are comparable to those reported from the United States,4,5 although some clinicians' experiences may include younger patients with disseminated gonococcal infection (DGI) or HIV infection or older patients with prosthetic joints.6 Patients with damaged joints are at increased risk. Infection complicates 0.05% to 0.48% of arthroscopies7; a higher risk is associated with reconstructive or repeat procedures. In rare cases, infection complicates arthrocentesis, intra-articular injection therapy, or direct traumatic penetration of the joint.
Presumably, most bacteria reach the joint via the bloodstream and the vascular synovial membrane (and, in children, the epiphyseal plate). Joint invasion may be facilitated by the absence of a developed basement membrane in the normal synovium. Previously damaged joints seem to be at increased risk for infection,1,6 which may be attributable to unique properties of Staphylococcus, the neovascularization of inflamed synovial membranes, or the increased expression of adhesion molecules on activated synovial endothelial cells. Some specificStaphylococcus, Streptococcus, and Neisseriae strains exhibit tropism to joints.8 In rare instances, joints are directly infected by traumatic inoculation; this circumstance may be more common in infections involving small finger joints.
Synovial reaction and biochemical changes to cartilage, attributable to mediators released from inflammatory cells and bacteria, occur within hours after experimental joint infection.9 Experimental models of bacterial arthritis have shown that even if antibiotic therapy is begun within 24 hours of infection, proteoglycans are already at least transiently depleted from cartilage. Loss of cartilage and erosion of subchondral bone may begin within days of infection and evolve into destruction of the joint. Polymorphonuclear neutrophils (PMNs), reacting to chemotactic factors, adhere to endothelial adhesion molecules and infiltrate the synovium, where they participate in the subsequent recruitment of mononuclear cells.
Some studies suggest that the degree of damage is determined by several factors, including the type of infecting organism and the accumulation of PMNs and their release of proteolytic enzymes and perhaps oxygen by-products into the joint space. N. gonorrhoeae induces only limited PMN activation in vitro, which may partially explain why minimal joint destruction is observed with this infection, even in studies of rabbits in which the bacteria is directly inoculated into knee joints. Preliminary animal studies suggest that potent anti-inflammatory therapy with corticosteroids, in conjunction with appropriate antimicrobial therapy, may reduce this cartilage destruction.10,11These observations have not yet been translated into clinical practice. Consistent with animal data, several studies reported that a delay in the initiation of treatment was a major determinant of poorer functional outcome.1
Diagnosis of Septic Arthritis
A thorough history remains a key element in the diagnosis of septic arthritis. Pertinent features include acute onset of joint pain or a significant change in the pattern of chronic joint pain; a history of penetrating joint trauma; a history of prodromal extra-articular symptoms suggestive of bacteremia; any comorbid immunosuppression, including diabetes mellitus, intravenous drug use, or prior intravenous catheterization; the presence of sexually transmitted diseases (STDs); and geographic location (e.g., in the case of Lyme disease). A complaint of sudden onset of articular or periarticular pain should never be ignored, especially if the pain is present both when the joint is at rest and when it is in motion. Because joint infections occur infrequently, patients and physicians may easily but incorrectly attribute the pain to trauma or overuse syndromes, particularly in the case of shoulder pain, or to crystal-induced arthritis if lower extremity or wrist synovitis is present.
Surprisingly, fever is not uniformly present in adults1,2,3,4,12 or children13 with septic arthritis. Fever may be present in fewer than 60% of patients with nongonococcal septic arthritis. Rigors may be present in fewer than 10% of patients.12 High spiking fevers (> 39° C [102.2° F]) and rigors can also occur in patients with crystal-induced arthritis.14 Thus, systemic features are neither sensitive enough nor specific enough to warrant making or excluding the diagnosis of septic arthritis without examination of the synovial fluid.
Detailed physical examination is warranted in all cases of possible septic arthritis, as in any patient with a possible systemic infection or inflammatory disorder. The pattern of joint involvement and the presence of tendinitis or enthesitis should be noted, as should any inflammation of the eyes, skin, or mucosae. The latter finding suggests the possibility of a reactive arthritis. Pain with passive motion of a joint or palpation of the joint capsule, in the absence of trauma, suggests synovitis. Mild to moderate swelling of an infected hip or shoulder, however, is usually not detectable by examination. Septic bursitis, most commonly affecting the prepatella or olecranon bursae,15must be distinguished from arthritis by clinical examination. Unnecessary arthrocentesis can be avoided by appropriate recognition of a periarticular bursitis that needs to be drained. Septic olecranon and prepatellar bursitis, which are the most common sites for septic bursitis, are often associated with a peculiar periarticular, often pitting, edema. A complete joint examination should be undertaken, with specific attention paid to the sternoclavicular, sacroiliac, and midfoot joints. In the absence of risk factors for osteomyelitis or stress fractures, exquisite tenderness of the midfoot should raise particular suspicion for gouty arthritis. Infection of fibrocartilagenous joints seems to be more common in intravenous drug users and in patients with indwelling lines. Psoriatic arthritis may produce dactylitis or mimic a septic joint; thus, even trivial psoriatic skin involvement should be noted. Potential portals of infection (e.g., prior I.V. access sites) should be sought out meticulously.
Synovial fluid leukocyte count, polarized microscopy, Gram stain, and culture are the most important initial laboratory investigations in the evaluation of suspected septic arthritis. Acutely or painfully swollen joints should be aspirated [see Figures 1 and 2] and the synovial fluid analyzed. Synovial fluid analysis is the diagnostic test of choice to distinguish between crystal-induced arthritis and infection-induced arthritis [see Differential Diagnosis of Acute Arthritis, below]. Testing for peripheral blood leukocytosis12,14 and neutrophilia, synovial fluid glucose measurement, mucin clot testing, and assessment for elevations in acute-phase reactants are of insufficient sensitivity and specificity to confirm or exclude the diagnosis of septic arthritis. Synovial fluid leukocytosis and neutrophilia are invariably present in septic arthritis, but their presence does not allow one to distinguish between infection-induced and crystal-induced arthritis [see 15:IX Crystal-Induced Joint Disease].16 Most infected joint fluids have a striking degree of leukocytosis. Although in approximately 25% of cases fluid from infected joints may have a white blood cell (WBC) count of less than 20,000/mm3, on differential counts, more than 85% of cells are almost always neutrophils.3,16 Infected bursal fluid may have only a few thousand WBCs. A rapid approximation of the WBC count can be obtained by microscopic examination of a drop of synovial fluid (magnification: × 40), with each WBC estimated to represent 500 WBCs/mm3.
Figure 1. Evaluation of Patients Based on Synovial Fluid Analysis
Algorithm for rapid evaluation of patients with acute monoarticular arthritis on the basis of synovial fluid analysis.
Figure 2. Approach to Patients with Monoarticular Arthritis
Approach for patients with persistent monoarticular arthritis on the basis of synovial fluid analysis.
Gram stain and culture of synovial fluid should be performed in all cases of suspected septic arthritis [see Table 1]. A Gram stain will permit evaluation of the percentage of PMNs, but it is not sensitive (< 60%) for the presence of bacteria.2,3,16 Gonococci are observed even less frequently. Antibiotic-containing culture plates should not be utilized, and samples must be delivered promptly to the laboratory. In the absence of prior antibiotic therapy, synovial fluid culture remains the gold standard for the diagnosis of nongonococcal joint infection. Synovial biopsy may have a slightly greater yield of positive cultures, but it is infrequently performed. There is insufficient evidence to recommend routine biopsy.
Table 1 Likely Causes of Septic Arthritis on the Basis of Gram Stain and Patient Demographics
Blood cultures should be obtained, as should cultures of any other potential extra-articular source of infection. If gonococcal infection is suspected, workup should include rectal, cervical, urethral, and pharyngeal cultures—any of which may be positive, even in the absence of local symptoms.17 Cultures and Gram stain of any suspicious skin lesions should be obtained, although these are usually sterile in cases of DGI.
A wet preparation of synovial fluid should be examined by polarized microscopy for the presence of crystals. In rare cases, crystal arthropathy and septic arthritis may coexist; thus, the presence of crystals does not rule out septic arthritis.5,18 Crystals can be found in synovial fluids from asymptomatic patients with a history of gout. If crystals are observed, patients are generally treated for crystalline arthritis alone [see 15:IX Crystal-Induced Joint Disease], unless there is real suspicion of coexistent infection. Hence, synovial fluid cultures should be obtained even when crystals are observed, especially if there is any suspicion of current or remote infection or if the use of therapeutic intra-articular corticosteroid injections is anticipated. Some clinicians order a bacterial culture for all inflammatory synovial fluids, even if crystals are observed. Local laboratory policy may dictate synovial fluid culture technique, but several studies suggest that the diagnostic yield may be improved with direct inoculation of fluid into blood culture vials or isolator tubes.
A potential advance in diagnostic testing has been the development of polymerase chain reaction (PCR) techniques for detection of bacterial DNA within synovial fluid and tissue.19 PCR has been used to diagnose the presence of Chlamydia and Yersinia organisms and the presence of Borrelia burgdorferi, N. gonorrhoeae, and Ureaplasma urealyticum DNA in synovial fluids. Caution must be used, however, when interpreting the results of PCR analyses. Contamination with even minute amounts of DNA can cause false positive results. The test also does not distinguish between the presence of live and dead organisms. Once PCR becomes better standardized and clinical experience is accumulated, PCR may be useful in diagnosing gonococcal, mycobacterial, and partially treated bacterial infections, which are not rapidly diagnosed by routine culture methods. The serum uric acid level should absolutely not be used to diagnose or exclude gouty arthritis; the level may be low, normal, or high at the time of a gout attack. Autoimmune serologies (e.g., antinuclear antibody and rheumatoid factor) are of no value in the initial management of patients with acute monoarticular arthritis. The erythrocyte sedimentation rate (ESR) or C-reactive protein (CRP) level may be helpful in monitoring response to treatment in some patients, particularly those with polyarticular or axial joint infections (in whom it can be difficult to assess reduction in local inflammation by physical examination). Although these approaches have not been tested in this specific setting, CRP has been shown to be of value in monitoring patients with bacterial endocarditis.20 The ESR may be normal in patients with septic arthritis, but it is usually elevated.1,2,12
In the absence of trauma, radiographs are of limited utility in the early diagnosis of acute synovitis. Radiographs are useful when there is clinical suspicion of chronic osteomyelitis, osteonecrosis, or pathologic or insufficiency fracture. Sequential radiographs are useful in evaluating a patient for the development of osteomyelitis. Ultrasound imaging is occasionally useful in documenting hip effusions in children or effusions in other joints in children or adults, especially joints having abnormal anatomy caused by prior mechanical or inflammatory trauma. Patients with suspected sternoclavicular or sacroiliac joint infections should have a computed tomography or magnetic resonance imaging scan performed to evaluate for mediastinal or pelvic extension. Because of its superior soft tissue imaging properties, MRI is useful for the evaluation of periarticular abscesses and soft tissue infections (e.g., psoas abscess mimicking septic arthritis of the hip and subdeltoid bursitis communicating with the shoulder joint). MRI can detect abnormal marrow signal secondary to osteomyelitis or avascular necrosis but may not help in distinguishing between septic and noninfectious inflammatory synovitis.21
Radionuclide scans and other imaging procedures are occasionally useful in localizing and defining the extent of infection. Radionuclide bone scans are sensitive early indicators of changes in periarticular bone metabolism. In a study comparing the use of a technetium-99 (99mTc) scan with the clinical gold standard of synovial fluid culture in patients with septic arthritis, 99mTc scanning had a sensitivity of 100%.22Radionuclide scans are not specific enough to establish a diagnosis. Scanning with 99mTc is helpful in differentiating bone from adjacent soft tissue inflammation.23 Gallium-67 (67Ga) localizes to areas of inflammation and increased bone metabolism. In cases of septic arthritis related to intravenous drug use, 67Ga scans showed positive results several days before 99mTc scans did. The 67Ga scan has been largely replaced by the indium-111 (111In) leukocyte scan. Although less sensitive than 99mTc, 111In is thought to be more specific for joint infection than either 99mTc or 67Ga, because leukocytes do not normally migrate to areas of increased bone metabolism associated with aseptic processes. No radionuclide scan will reliably distinguish between infection-induced and crystal-induced arthritis.
In the vast majority of patients with septic arthritis, especially those from whom synovial fluid can be obtained, radionuclide imaging studies are unnecessary.
Differential Diagnosis of Acute Arthritis
In the patient with acute monoarticular or oligoarticular arthritis, the clinician must promptly distinguish between crystal-induced arthritis and infection-induced arthritis, because the initial treatments of these two conditions are strikingly different. The diagnostic test of choice is synovial fluid analysis. Acutely inflamed or swollen, painful joints, in the absence of obvious trauma, should be aspirated. A single drop of fluid (even the small amount in the hub of the needle of an apparently unsuccessful arthrocentesis) may be sufficient to establish a diagnosis of crystal-induced arthritis. Wet preparation can provide an initial cell count estimate and a polarized microscopic evaluation for crystals. The fluid sample should then be air-dried and a Gram stain performed to both screen for bacteria and calculate the percentage of PMNs. The inability to demonstrate eosinophils by use of this stain technique is rarely of importance. Reactive and enteropathic arthritis are diagnoses that should generally be made only after infection has been excluded by culture of synovial fluid.
Specific Infectious Agents
Causative pathogens are documented by culture in approximately two thirds of presumed septic arthritis cases in most series, and identification rates as high as 93% have been reported. A portal of entry can be identified in half of the patients with septic arthritis; approximately 25% of cases are iatrogenic. In most series, 75% of cases have a possible extra-articular source of infection (e.g., cutaneous, respiratory, or genitourinary).
Determining the causative organism has implications for initial and adjuvant therapies, especially if the organism is resistant to particular therapeutic agents.
Traditionally, bacterial arthritis is divided into nongonococcal and gonococcal types. The distinction is clinically useful because gonococcal infections have a better prognosis than nongonococcal infections. Viral arthritis, which tends to have a self-limited course (except in some cases induced by parvovirus and hepatitis B or C), will not be discussed in this chapter.
In some practice settings, DGI is a relatively common cause of septic arthritis and tenosynovitis in healthy, sexually active patients. In a review of 41 cases of gonococcal arthritis, 83% of patients were female, and the mean age was 23 years.17 Knee synovitis was present in 54% of the patients, followed in frequency by hand and wrist synovitis. Involvement of the hip, but not the spine, is common. Dermatitis (usually sparse peripheral necrotic pustules) and migratory polyarthralgias/polyarthritis were present in 39% and 66% of the patients, respectively. Along with tenosynovitis, these findings constitute the classic triad of DGI. Genitourinary involvement was noted in 63% of the patients but is often asymptomatic in women. Absence of pelvic inflammatory disease does not exclude a diagnosis of DGI. Rectal and pharyngeal colonization are commonly asymptomatic; these sites, as well as blood, should be routinely cultured in all patients suspected of having DGI. Female genital infection may have occurred long before systemic dissemination. The frequency of positive cultures in one study of DGI was as follows: urogenital, 86%; synovial fluid, 44%; rectal, 39%; blood, 13%; and pharyngeal, 7%.17
Arthritis and vasculitic skin lesions may reflect local metastatic infection or a sterile inflammation induced by immune complexes. This dual pathophysiology may partially account for both the low frequency of positive joint cultures and the favorable joint outcome. Data compiled by the Centers for Disease Control and Prevention indicate that up to 30% of gonococcal isolates obtained in STD clinics in the United States in 1997 were resistant to penicillin or tetracycline.24 Resistant strains are capable of systemic dissemination. There have been only rare reports of resistance to ceftriaxone; therefore, initial therapy should include parenteral therapy with ceftriaxone or ciprofloxacin. Fluoroquinolone resistance is now rarely documented.24 Appropriate regimens, once the infection has been identified, include ceftriaxone (1 to 2 g intramuscularly or intravenously once daily), imipenem (0.5 g intravenously every 6 hours), or cefotaxime (1 g intravenously every 8 hours until 24 to 48 hours after resolution of symptoms). If DGI is clinically suspected but no organism is cultured, empirical therapy with a broad-spectrum cephalosporin is generally continued for approximately 1 week until the lack of any response is documented. Subsequent outpatient therapy should be continued with ciprofloxacin (500 mg orally b.i.d.) or ampicillin (500 to 1,000 mg orally q.i.d.), if the identified pathogen shows sensitivity to these agents. Recognized infection should always prompt an evaluation for other STDs, including syphilis and HIV. Empirical treatment for Chlamydia trachomatis infection (doxycycline, 100 mg orally b.i.d. for 7 days, or azithromycin, 1 g orally in a single dose) should also be given because this infection is frequently asymptomatic and can result in infertility if untreated; both partners should be treated whenever possible. Dosages may need to be modified in the presence of severe renal insufficiency. Once the diagnosis is confirmed and clinical response exhibited, addition of an NSAID is a reasonable adjunctive therapy because, as noted above, the synovitis may be induced by immune complexes as well as direct infection.
A dermatitis-arthritis syndrome, similar to DGI, can also be associated with Haemophilus influenzae, N. meningitidis, and Streptobacillus moniliformis infections, as well as with endovascular infections. Disseminated Neisseria infections, which may be recurrent, have been associated with the presence of terminal complement deficiencies.
Gram-positive bacteria remain the most common cause of septic arthritis, accounting for 70% to 80% of cases. S. aureus accounts for more than half of the cases of culture-positive septic arthritis in studies at university hospitals and for even higher percentages of certain patient subgroups: 70% to 80% of patients with polyarticular septic arthritis; more than 80% of infected patients with rheumatoid arthritis (RA)25; and 82% of infected hemodialysis patients.26 Staphylococcal arthritis was particularly frequent in a series of patients with endocarditis related to intravenous drug abuse.27 The predominance of S. aureus in septic arthritis has remained unchanged over the past 40 years,1 and there is now an increasing prevalence of methicillin-resistant S. aureus (MRSA) and S. epidermidis in patients with prosthetic joints. S. aureus should always be considered when selecting antibiotics in the initial treatment of presumed septic arthritis.
Gram stain cannot be relied on to differentiate between Staphylococcus and Streptococcus, because in biologic smears, Staphylococcus may not exhibit the clusters seen when grown in vitro. Bacteremia should be assumed in cases of staphylococcal arthritis, although it cannot always be documented. Endovascular infection with possible metastatic seeding to other areas should be considered in patients with persistent fever, leukocytosis, thrombocytosis, or an elevated ESR or CRP level.
Suspected staphylococcal joint infection should be treated initially with vancomycin (1 g intravenously every 12 hours, if renal function is normal) until methicillin resistance can be excluded. An appropriate penicillin or a cephalosporin can often be substituted for completion of a 4- to 6-week treatment course once the sensitivity to these agents is identified.
Non-group A, β-hemolytic streptococci are the second most common cause of septic arthritis, accounting for 10% to 21% of culture-positive cases.1,2,3 The number of reported group B (and to a lesser extent, groups C and G) streptococcal infections has been increasing.28 Group B streptococcal infection may be particularly virulent in diabetic patients; it involves axial joints (sacroiliac, sternoclavicular, and manubriosternal) and may be associated with poor functional outcome. Other manifestations of group B streptococcal sepsis include myositis, fasciitis, and endophthalmitis. Some investigators have expressed concern over the development of tolerance to penicillin in these usually penicillin-sensitive organisms and have suggested adding an aminoglycoside to therapy or using alternative antibiotics until sensitivity and bactericidal tests are available. S. pneumoniae may account for up to 10% of the cases of septic arthritis. In one study, 11% of patients with septic arthritis requiring transfer to the intensive care unit had pneumococcal infection.5 The incidence of penicillin-resistant streptococci, including S. pneumoniae, is rising. Polyarticular pneumococcal infections occur, which is not surprising given the high incidence of bacteremia in patients with pneumococcal pneumonia. Elderly adults may be particularly susceptible, and fever or an extra-articular source of infection may not be evident.29 Antibiotic therapy must be tailored with knowledge of local resistance patterns. Enterococci can occasionally infect native or prosthetic joints.30
Diagnosis is made on the basis of a Gram stain or preliminary culture showing gram-positive cocci; for initial therapy, vancomycin (1 g every 12 hours, with normal renal function) is a reasonable choice. Definitive therapy should be determined on the basis of culture results.
Gram-negative organisms account for 9% to 20% of septic arthritis cases reported at teaching hospitals and occur most frequently in immunosuppressed patients.1,2,3,4 Although certain pathogens have been associated with specific patient subgroups (e.g., Salmonella in patients with systemic lupus erythematosus or sickle cell anemia, Pseudomonas aeruginosa in intravenous drug abusers), joint infections with these organisms can occur in immunocompetent patients. In sickle cell patients, Salmonella more commonly causes osteomyelitis than septic arthritis and may infect multiple sites.31
Comorbid medical conditions, prior antibiotic use, and extra-articular infections (particularly urinary tract infections and decubitus ulcers) predispose to gram-negative septic arthritis. Reviews of septic arthritis in intravenous drug abusers over the past decade report a lower percentage of gram-negative pathogens than were reported in earlier series, in which organisms such as Enterobacter, Klebsiella, Serratia, and P. aeruginosa accounted for more than 80% of infections.27 Pathogens and sensitivities associated with intravenous drug abuse are specific to geographic regions, and choice of antimicrobial therapy for this patient subgroup must take these factors into account. This same population is also at risk for HIV and viral hepatitis.
Ampicillin is bactericidal for Salmonella; however, an increasing number of Salmonella infections are resistant to this drug. Chloramphenicol, trimethoprim-sulfamethoxazole, and especially quinolones are used as alternative agents.31 For other gram-negative infections, cefotaxime (2 g intravenously every 6 hours) or imipenem (0.5 g intravenously every 6 hours) can be used with or without an aminoglycoside until culture results arrive. If pseudomonal infection is suspected, it is reasonable to initiate dual drug therapy, including an aminoglycoside.
Anaerobic joint infections are rare, accounting for at most 7% of cases in reported series; however, they should be considered in patients with deep-tissue or abdominal abscesses. Intra-articular emphysema has been described with anaerobic infection (Clostridium species) but is not specific to anaerobes.32 Empirical treatment for anaerobic septic arthritis is generally not undertaken unless there is a coincident infection with a high likelihood of anaerobic involvement. In those cases, treatment should be based on the suspected organisms and the portal of infection (oral, abdominal, or traumatic).
This destructive U. urealyticum septic arthritis should be distinguished from the rheumatoid-like, but generally nonerosive, sterile polyarthritis associated with hypogammaglobulinemia. Initial treatment of acute-onset arthritis in this setting should include both a tetracycline (e.g., doxycycline, 100 mg every 12 hours) and aggressive intravenous immunoglobulin replacement until Ureaplasma infection is excluded.
Pasteurella multocida, a pleomorphic gram-negative coccobacillus parasitic to many cats and dogs, is an infrequent cause of septic arthritis.35 Infection generally follows inoculation from an animal bite, although in one third of cases no trauma is documented. The organism is almost always sensitive to a penicillin; aminoglycosides are generally ineffective. Initial antibiotic coverage after a recent animal bite should include a penicillin or a third-generation cephalosporin. Ampicillin-sulbactam (1.5 g intravenously every 6 hours) and ceftriaxone (2 g intravenously q.d.) are reasonable initial therapies pending final culture results.
Mycobacterium marinum is an atypical mycobacterium acquired through exposure to freshwater, saltwater, swimming pools, or fish tanks. M. marinum usually causes cutaneous and subcutaneous infection, but it can cause subcutaneous nodules, tenosynovitis, and septic arthritis.36,37 Monoarticular, often proliferative, involvement of the metacarpophalangeal and proximal interphalangeal joints is most frequently reported. The arthritis may be only slightly painful with no systemic symptoms. Delays in diagnosis are common (mean, 8 months). Optimal growth of the organism from synovial fluid or tissue requires special attention in the microbiology laboratory. M. marinuminfection may elicit a positive purified protein derivative (PPD) reaction. Treatment usually consists of prolonged chemotherapy and surgical debridement. There are no evidence-based guidelines to aid in determining the need for initial surgical debridement, although this procedure is frequently utilized, especially when the course of infection is protracted. A reasonable initial therapy is a combination of rifampin (600 mg orally) and ethambutol (15 mg/kg/day; maximum, 1,600 daily), although treatment with trimethoprim-sulfamethoxazole (160/800 mg orally b.i.d.) or ciprofloxacin (500 to 750 mg orally b.i.d.) may also be effective. Although there are no data demonstrating the superiority of combination therapy, there has been anecdotal treatment failure with doxycycline monotherapy. The organisms are usually sensitive to amikacin and clarithromycin.
Arthritis is an uncommon extrapulmonary complication of tuberculosis38,39; osteomyelitis of the low thoracic and lumbar spine occurs more frequently.40 Recognition of tuberculous peripheral joint and bursa infections are often significantly delayed, partly because of the indolent nature of the inflammation, the rarity of the disease, and the frequent absence of obvious pulmonary disease or a positive PPD test.38Although joint motion is frequently maintained, radiographs may show destruction of bone and loss of cartilage. Synovial fluid leukocytosis is modest to marked, usually with a predominance of neutrophils. Acid-fast staining of fluid is usually negative; cultures are usually positive. Limited data suggest that biopsy of synovium may yield a higher percentage of positive cultures than do cultures of fluid. BecauseM. tuberculosis is a rare cause of peripheral arthritis, it is not cost-effective to obtain mycobacterial cultures initially from all patients with acute synovitis. Microbiologic diagnosis takes several weeks; therefore, molecular diagnostic tests for M. tuberculosis DNA should be requested only if there is a significant clinical concern about this infection. No large studies have been undertaken to compare diagnostic approaches or treatment regimens for tuberculous arthritis. As with therapy for pulmonary tuberculosis, there is a potential for developing drug resistance; thus, initial therapy should include a combination of at least two drugs, including rifampin (600 mg/day), isoniazid (300 mg/day), and perhaps ethambutol (15 mg/kg/day) for the first few months of a planned 18- to 24-month course of treatment.
The arthritis associated with later stages of syphilis is characteristically a chronic, proliferative, symmetrical polyarthritis involving small and large joints.41 An acute monoarthritis is not typical of syphilis. Synovial fluid usually contains a predominance of mononuclear cells. Treatment should be based on the stage of disease (usually secondary or tertiary) and the presence or absence of HIV infection [see 7:VI Syphilis and the Nonvenereal Treponematoses]. Penicillin remains the antibiotic of choice, although ceftriaxone or doxycycline may be used in some patients without HIV infection.
Fungal arthritis is rare and is seemingly more common in immunosuppressed patients. The arthritis is usually subacute or chronic. Osteomyelitis is more common than arthritis in Cryptococcus, Coccidioides, and blastomycosis infections.42,43 Arthritis associated with histoplasmosis may occur with erythema nodosum and may clinically mimic sarcoidosis. Arthritis caused by aspergillosis infection has been reported after systemic infection and iatrogenic inoculation at the time of arthrocentesis. Sporotrichosis arthritis can occur after cutaneous inoculation or, in rare cases, through inhalation.44 Surgical debridement may be helpful; there are limited data, however, to support initial surgery in addition to antifungal therapy. Except in cases of local inoculation (e.g., rose thorn sporotrichosis), the fungal infection is generally assumed to be systemic, and intravenous amphotericin B should be administered. Intra-articular amphotericin B and oral regimens with azole antibiotics have been utilized. These alternative agents and routes of administration should be considered on an individual basis; consultation with an infectious disease specialist should take place early in the treatment program.
Clinical Subgroups at Risk for Septic Arthritis
Several factors predispose certain groups of patients to septic arthritis. Local risk factors include the presence of inflammatory arthritis, prior intra-articular corticosteroid injection, and the presence of implanted synthetic material.6 Other factors include advanced age, a remote focus of infection, diabetes mellitus, indwelling lines, intravenous drug use, and immunosuppression.
Age older than 60 years has been repeatedly found to be a risk factor for septic arthritis, perhaps in part because of comorbid conditions.1,6Prosthetic joints, which also pose an increased risk of infection, are more common in older patients.
The clinical presentation of septic arthritis tends to be more insidious in elderly patients than in younger patients. The absence of fever is common.29 Older patients are particularly at risk for septic arthritis as a postoperative complication, often after a bacteremic episode. In a retrospective study of 21 patients older than 60 years with septic arthritis, mortality was 21%.2
RHEUMATOID ARTHRITIS PATIENTS
Patients with RA have consistently accounted for a disproportionate percentage of patients with septic arthritis; in some series, 50% of septic arthritis patients had RA. These patients may be predisposed to septic arthritis because of poor clearance of bacteria from abnormal joints or because of phagocytic defects acquired secondary to drugs or disease. Patients with long-standing, severe RA are more likely to develop septic arthritis than patients with less severe disease. The majority of RA patients with septic arthritis present with insidious worsening of joint symptoms, and the diagnosis of infection is often delayed.25 Although there have been reports of so-called pseudoseptic rheumatoid joints characterized by highly inflammatory, sterile synovial fluids that improve with antirheumatic therapy, any isolated, acutely inflamed joint in a patient with RA should be considered infected until proved otherwise.
With septic arthritis, the outcome is significantly worse for RA patients than nonrheumatoid patients, perhaps because of delays in diagnosis. Reported mortality for RA patients with monoarticular infection has ranged from 15% to 22%. When both monoarticular and polyarticular infections are included, mortality reaches 44% in RA patients, compared with 26% in unselected septic arthritis patients. For patients with a combination of polyarticular septic arthritis and RA, mortality is 50%. Joint outcome is also poor; only 35% of RA patients achieve full functional recovery of the infected joint, compared with 70% of nonrheumatoid patients. A retrospective literature review of 181 RA patients reported recurrence of infection in nonprosthetic joints in 20% of patients.25
INTRAVENOUS DRUG ABUSERS
Intravenous drug abusers are at risk for repetitive transient bacteremia and direct bacterial inoculation of soft tissue. Bacteremia is thought to be the most common mechanism of infection. Although medium to large joints (such as the wrist and knee) are most commonly involved, fibrocartilaginous joints (sternoclavicular, sacroiliac) and the spine are frequently infected. Gram-positive organisms (especially S. aureusand MRSA) are the most common pathogens,27 followed by gram-negative bacteria (Enterobacter, Serratia, and P. aeruginosa).
Not all studies of septic arthritis in intravenous drug abusers have specified whether the patients were also HIV positive. Two Spanish studies, one with 25 patients and one with 16 patients, reported no differences in clinical characteristics, infecting organisms, or outcome between intravenous drug abusers with HIV and those without HIV.45
Hemodialysis patients are at increased risk for septic arthritis because of recurrent vascular infections, intravenous catheterizations, repeated skin trauma, and immune deficits such as decreased clearance of transient bacteremia. Hemodialysis patients are at particularly high risk for S. aureus colonization, with reported nasal carriage rates over 40%.
A 10-year review of hemodialysis patients in three tertiary care hospitals identified 11 episodes of septic arthritis, with 82% attributable toS. aureus.26 Almost all of the episodes involved joints above the diaphragm, including one sternoclavicular and one acromioclavicular joint infection.
Two articular syndromes unique to HIV-infected patients have been reported that can mimic acute septic arthritis. The first, which has been referred to as AIDS-associated arthritis, is characterized by the development over several days of extreme pain and disability involving the knees or ankles, noninflammatory sterile synovial fluid, and an excellent response to nonsteroidal anti-inflammatory drugs (NSAIDs). The reported duration of symptoms is 1 to 6 weeks. The second syndrome, the painful articular syndrome, is characterized by an explosive onset of pain in the shoulders and elbows. The pain resembles gout in its intensity and often requires intravenous narcotics for analgesia. Synovial fluid is noninflammatory, and symptoms last a few hours to several days. The etiology of these syndromes remains unknown.
Despite the multiple immune deficits associated with HIV infection, there have been relatively few reported cases of septic arthritis in patients with HIV. S. aureus and S. pneumoniae are the most commonly reported organisms.46 With an increasing degree of CD4+ T cell depletion, infection with Candida albicans, Cryptococcus neoformans, Histoplasma capsulatum, Nocardia asteroides, and Sporothrix schenckiican occur. HIV patients are also at increased risk for pyomyositis, most commonly occurring secondary to S. aureus.46
ORGAN TRANSPLANTATION PATIENTS
A review of the University of Pennsylvania renal transplant experience reported the occurrence of septic arthritis in 0.8% of the patient population.47 All cases occurred within 18 months of transplantation, the period of most intense immunosuppression with high-dose steroids. Infection was caused by bacteria in 66% of cases (half of which were caused by S. aureus), mycobacteria in 17%, and disseminated cytomegalovirus infection in 17%. Only three cases of fungal arthritis have been reported in renal transplant patients, but this incidence is likely underreported. With the current ubiquitous use of cyclosporine, gouty arthritis (occasionally polyarticular with atypical joint involvement) is extremely common and may mimic septic arthritis.48
INDIVIDUALS WITH PROSTHETIC JOINTS
Prosthetic joint infections, in the absence of wound drainage, are frequently not recognized early and may result in a nonfunctional joint with chronic osteomyelitis. Pain is the predominant or only symptom.49 Effusions may be small (knee) or impossible to detect clinically (hip or shoulder). Fever and leukocytosis are frequently absent. In studies, S. epidermidis was responsible for approximately 40% of both early postoperative and late infections.49 Although the ESR is usually elevated, it may be normal.50,51 One study demonstrated a sensitivity of only 60% when an ESR threshold greater than 30 mm/hr was used.50 The CRP assay may be a more sensitive test.51 Three-phase bone scans are also surprisingly insensitive.51 If loosening is noted on standard radiographs, bone scanning is of little additional diagnostic value. Arthrocentesis or intraoperative tissue sampling with culture, before antibiotic therapy, remains the gold standard for diagnosis. The value of fluid Gram stain is questionable.51 Knee replacement can be repeated in some patients as a two-step procedure. Total debridement, followed by 4 to 6 weeks of systemic antibiotic therapy before attempted repeat arthroplasty, is frequently successful.52
Treatment of Septic Arthritis
Treatment of septic arthritis consists of drainage, parenteral antibiotics (guided by Gram stain, if possible [see Table 1]), and initial (not prolonged) joint immobilization for pain control. The controversy surrounding medical versus surgical drainage remains unresolved. A retrospective analysis by Broy and Schmid included pooled data from 80 studies.53 A slightly larger number of patients who were treated medically (needle drainage) rather than surgically (arthrotomy or arthroscopy) had a good joint outcome (66% versus 57% [P < 0.05]) with restoration of normal joint function and minimal residual pain. Perhaps because of increased systemic comorbidity in the medically treated cohort—an inherent problem in nonrandomized studies—medically treated patients had a higher mortality (10% versus 3%; P < 0.05). These nonrandomized studies do not permit generation of firm therapeutic guidelines.
Clinicians generally agree that surgical drainage is indicated in the following situations: septic arthritis of the hip or of joints that are difficult to aspirate or monitor for adequate drainage, extensive spread of infection to the soft tissues, and inadequate clinical response to appropriate antibiotics after 5 to 7 days. There is an apparent consensus among rheumatologists that surgical drainage is rarely necessary for patients with gonococcal septic arthritis in whom the response to appropriate antibiotics is characteristically rapid (less than a week) and complete. No prospective or controlled studies support or refute the need to openly drain or lavage an uncomplicated, acutely infected, nongonococcal joint at the initiation of therapy; nonetheless, this orthopedic practice is common in many parts of the country. Surgical drainage should be considered when a previously damaged (e.g., rheumatoid) joint becomes infected, because the data indicate a particularly poor outcome for these patients with medical therapy.25 The data are insufficient to mandate immediate surgical drainage of infected joints in immunosuppressed patients. In general practice, arthroscopic lavage seems to be supplanting open joint drainage; it permits earlier physical therapy and mobilization of the involved joint.54
The choice of antibiotics [see Table 2] should be guided by patient characteristics and the local microbiologic sensitivity pattern. Intra-articular antibiotics are not required55 and may cause a chemical synovitis. Parenteral antibiotics are generally administered for approximately 4 to 6 weeks. Acetaminophen and NSAIDs should be avoided until the diagnosis is solid (i.e., through microbiologic identification or by documented defervescence and clinical improvement with antibiotics alone). High-dose NSAIDs can transiently blunt the inflammatory response and may suggest a response to coadministered antibiotics, which might then be construed as evidence for a culture-negative (e.g., gonococcal) infection. There are no data demonstrating that NSAIDs prolong the course of infection, and some animal data suggest that anti-inflammatory therapy may be beneficial.10,11 Narcotics may be used initially for pain control. Joint immobilization for a short period (1 to 3 days) is appropriate to help with pain control, but daily range-of-motion exercises should be started as soon as possible to hasten return of joint function.56 Once a definitive diagnosis is made, and if there are no contraindications, NSAID therapy may be initiated to facilitate joint mobilization.
Table 2 Initial Empirical Antibiotic Therapy for Acute Septic Arthritis in Adults*
The author thanks Dr. Mathilde Pioro for her assistance in the preparation of an initial draft of the manuscript.
Editors: Dale, David C.; Federman, Daniel D.