Chapter 48. Septic Arthritis

Septic, pyogenic, and suppurative arthritis are the names given to the inflammation of the joint space caused by the presence of bacteria or fungi. Septic arthritis is more common in childhood than in any other period of human life and more than half of cases are diagnosed in individuals younger than 20 years of age. Since septic arthritis usually has a hematogenous origin, the age distribution of pediatric patients with joint infection is markedly skewed, reflecting the increased attack rate of bacteremia in early childhood. In a large series of 725 pediatric patients with joint infections compiled by Trujillo and Nelson, 52% of the children were younger than 2 years, 25% were aged 2–5 years, 15% were 6–10 years old, and the remaining 6% were aged 11–15 years.1 Since a significant fraction of suspected cases of septic arthritis remains bacteriologically unconfirmed, the true incidence of the disease is uncertain. The estimate annual incidence of the disease in the general population ranges between 2 and 10 cases per 100,000.2 Several pediatric subpopulations are at increased risk for septic arthritis, as summarized in Table 48–1.

The highly vascular synovial tissue lacks a limiting basement membrane, enabling easy access of circulating bacteria to the joint space during an episode of bacteremia. Once organisms have penetrated into the joint, the low fluid shear conditions facilitate microbial adherence. Occasionally, septic arthritis results by direct inoculation of bacteria in the joint by penetrating trauma, bites, intra-articular injections (particularly corticosteroids) or a surgical procedure. In neonates and young infants, bacteria may migrate from an adjacent focus of osteomyelitis into the joint traversing through capillaries that cross the metaphyseal growth plate. This capillary network recedes between 6 and 9 months of age and in the older child only the metaphyses of the hip, shoulder, and ankle bones remain intracapsular.3

A variety of bacterial adhesins have been implicated in anchoring organisms to the synovial layer, which explains the virulence and tropism exhibited by bacteria such as Staphylococcus aureus, Streptococcus agalactiae (group B), Nesisseria gonorrhoeae, and Borrelia burgdorferi. These adherence-promoting molecules, termed microbial surface components recognizing adhesive matrix molecules, have been best studied in S. aureus and include, among others, fibrinogen-, fibronectin-, and elastin-binding proteins, a collagen receptor, and an adhesin with wide specificity. Mutations in the genes encoding for these proteins markedly reduce or abolish the capability of the organism to cause septic arthritis in animal models.3

Trauma may facilitate the entrance of circulating bacteria into the joint space caused by increased local vascularization, whereas high concentration of a diversity of protein fibers in the synovial fluid after surgery may promote bacterial adherence to the tissue. The important role played by antecedent trauma is well exemplified by the propensity of S. agalactiae to invade the shoulder in newborns delivered in cervical presentation, and the occurrence of septic arthritis of the hip among those delivered in breech presentation.

In a minority of cases the host's immune system is able to contain the infection and this may account for transient arthralgias and arthritis observed in some patients with bacteremia caused by pathogens of low virulence such as Kingella kingae.4,5 In most cases, however, the infection progresses and organisms multiply in the joint space to elicit an acute inflammatory response.

Both bacterial virulence factors and the host immune response appear to contribute to the progressive destruction of joint architecture in patients with septic arthritis. Following adherence, organisms such as staphylococci may be internalized into osteoblasts through membrane pseudopod formation or induction of endocytosis. Internalized bacteria may induce apoptosis or avoid the immune system defenses by surviving and multiplying inside the cell. As organisms multiply in the joint, a sequence of proliferation of the synovial lining cells, mononuclear-cell infiltration, granulation tissue, and abscess formation occurs.3 Eventually, cartilage and bone destruction ensue as a result of release of leukocyte proteases, inflammatory cytokines (especially interleukin (IL)-1, IL6, and tumor necrosis factor-α), and bacterial toxins, as well as tissue ischemia and necrosis induced by increased intra-articular pressure.3 Over time, cartilage degradation causes narrowing of the joint space and further erosive damage, leaving long-term orthopedic disabilities.

In the preantibiotic era, S. aureus, β-hemolytic streptococci and Streptococcus pneumoniae were the most common organisms identified in pediatric patients with septic arthritis.6 After the introduction of effective antimicrobial therapy, the incidence of streptococci declined and S. aureus remained the predominant etiology of skeletal system infections. In the 1960s, implementation of routine seeding of synovial fluid aspirates onto chocolate-agar plates, resulted in the recognition of Haemophilus influenzae type b as the most common cause of suppurative arthritis in young children and emphasized the crucial role played by the adequate culture techniques in the diagnosis of the disease. However, use of multiple solid media, for culturing synovial fluid aspirates obtained from children with presumptive joint infections yield the causative organism in only two-thirds of cases. Obviously, an incorrect diagnosis or previous antibiotic therapy may be responsible for a fraction of “culture-negative” septic arthritis cases. Yet, the possibility remains that some of these patients have a joint infection caused by fastidious organisms that are not detected by routine laboratory techniques.

In the late 1980s, inoculation of joint exudates into pediatric aerobic BACTEC™ blood culture vials (Becton Dickinson, Cockeysville, MD) resulted in the detection of K. kingae, a gram-negative commensal bacterium of the pharyngeal flora, as the most common pathogen of septic arthritis in young children, causing 48% of cases with a culture-proven etiology.7 Attempts to isolate the organism from joint or bone exudates on routine solid media failed in most cases, although subculture of positive blood culture vials onto blood-agar or chocolate-agar plates recovered K. kingae without difficulties, demonstrating that routine solid media are able to support its nutritional requirements. It is postulated that pus exerts an inhibitory effect upon K. kingae and dilution of exudates in a large broth volume decreases the concentration of detrimental factors, improving recovery of the organism.

In recent years, use of conventional and real-time polymerase chain reaction (PCR) technology and DNA sequencing has improved detection of difficult-to-culture organisms, enabling recognition of bacteria in patients already treated with antibiotics, and reducing the time-to-detection of the pathogen. Pioneer use of these novel approaches has confirmed that K. kingae is the most common etiology of septic arthritis in children below the age of 3 years and shortened the time required to detect and identify the causative organism from a few days to less than 24 hours.8–10 In a recent study by Chometon et al. that included 131 children with presumptive septic arthritis, a bacterial pathogen was identified by culture in 59 (45%) specimens, of which 25 grew S. aureus and 17 (29%) grew K. kingae.11 The combination of culture, conventional PCR followed by sequencing of the amplicon, and real-time PCR with K. kingae-specific probes increased the overall bacterial identification to 61% and the fraction of specimens in which K. kingae was detected to 30%.11 It is to be expected that further development and accumulative experience with nucleic acid amplification methods will substantially reduce the frustrating proportion of cases of “culture-negative septic arthritis” and contribute to a better management of these patients.

Generally, septic arthritis in children is caused by the intra-articular invasion of a single bacterial species. Isolation of multiple organisms should raise the suspicion of immunodeficiency, drug abuse, or penetrating trauma with direct inoculation of bacteria into the joint space.

The etiologic agents of septic arthritis show a clear age-related distribution (Table 48–2). S. aureus is the most common cause of joint infections in neonates and children older than 2–3 years. In recent years, methicillin-resistant strains of S. aureus (MRSA) are being detected in the community among patients lacking the established risk factors for nosocomial MRSA infections.12 In some areas, an increase in the incidence and severity of skeletal system infections is being noted coinciding with the emergence of these strains, whereas the rate of infection caused by methicillin-susceptible S. aureus and other organisms remain stable.13 Patients infected with community associated MRSA (CA-MRSA) organisms are usually healthy children and adolescents frequently involved in contact sports. Infecting strains harbor a unique staphylococcal chromosome cassette termed SCCmec type IV (and less commonly SCCmec typeV) that carries fewer antibiotic resistant determinants compared to MRSA isolates of nosocomial origin. Infections with CA-MRSA usually involve skin, soft tissues, the lung, and the skeletal system and are characterized by remarkable tissue destruction. More than 90% of CA-MRSA isolates possess the Panton-Valentine leucocidin, which is rarely detected in methicillin-susceptible organisms. It is unclear yet whether this toxin is the major virulence factor of CA-MRSA organisms or is merely a biological maker for some other and still unidentified bacterial component.14

S. agalactiae and gram-negative enteric bacilli are seen almost exclusively in the neonatal period. S. pneumoniae is most common between the ages of 6 months and 2 years, reflecting the increased incidence of bacteremia and invasive diseases caused by encapsulated bacteria in young children with physiological immaturity of the T-cell independent immunity.15,16Streptococcus pyogenes is isolated 10–20% of children with septic arthritis regardless of age and is especially common in patients with concomitant skin infections and varicella.17

H. influenzae type b was the most common cause of invasive disease in children younger than 2 years prior to the advent of the conjugated vaccine, accounting for almost one-half of the cases. The disease has become rare in countries where immunization coverage is high.18 Children with H. influenzae arthritis frequently present with other suppurative foci of infection such as meningitis (in 30% of patients), osteomyelitis (in 22%), cellulitis (in 30%), pneumonia (in 4%), and otitis media (in 35%).19

More than 90% of all children with K. kingae arthritis are 6–30 months of age.20 Antecedent or concomitant stomatitis—including that caused by primary herpetic infection—or signs of an upper respiratory tract infection are common, suggesting that invasion of the bloodstream by organisms carried in the pharynx is facilitated by mucosal breaching.

Pseudomonas aeruginosa is a rare cause of septic arthritis in the general population but it may cause infection in neonates and drug abusing adolescents.21

N. gonorrhoeae becomes common in sexually-active adolescents and its isolation in young children is a marker of sexual abuse. Gonococci may also invade the joint space in the course of a disseminated disease in neonates born to infected mothers.22 Although the incidence of arthritis in invasive meningococcal disease is as high as 14%, true invasion of the joint space by Neisseria meningitidis is uncommon.23 In most cases, articular involvement in patients with meningococcemia develops several days after initiation of antibiotic therapy and the synovial fluid is usually sterile. Immunocomplexes have been detected in some of these patients, suggesting a reactive nature. Recurrent disease, a prolonged course, isolation of uncommon N. meningitidis serogroups, and family clustering of cases should raise the possibility of complement or properdin deficiencies.24

Invasion of the joint space by Salmonella enterica has been rarely reported, usually affecting infants and young children frequently suffering from sickle cell anemia and other hemoglobinopathies.25 In the developed world, as the result of effective public health measures, human brucellosis has been practically eradicated. In children with arthritis, residents, of endemic countries (Latin America, the Middle East, the Mediterranean basin, Eastern Europe, Asia, and Africa), and travelers returning from these regions, the possibility of brucellar arthritis (and especially that caused by Brucella melitensis) should be entertained.26

Lyme disease should be included in the differential diagnosis of children exposed to ticks in areas where the infection is prevalent who present with joint inflammation. Septic arthritis caused by Mycoplasma and Ureaplasma species is almost exclusively detected in patients with X-linked agammaglobulinemia, common variable immunodeficiency, or organ transplantation.27

Tuberculous septic arthritis is usually monoarticular and generally affects the hips or knees. Invasion of the joint space occurs when Mycobacterium tuberculosis bacilli cross the epiphyseal plate from a contiguous tuberculous bone lesion or more rarely by the direct hematogenous seeding of the organism into the synovium.28 The lack of proteolytic enzymes preserves the joint space in early disease, but cold abscesses and draining sinus tracts may occur in long-standing, neglected disease.

Hematogenous septic arthritis caused by anaerobic organisms is exceptionally seen in children and is usually caused by a single bacterial species, generally a gram-negative bacillus. Whenever a penetrating wound or bite is the mechanism of infection, multiple organisms including both aerobes and anaerobes, may be isolated from the joint fluid culture.29

Rat-bite fever is a rare zoonosis caused by two members of rodents’ oral flora Streptobacillus moniliformis, mostly in Western countries and Australia, and by Spirilum minus in Asia.30 The site of inoculation of the disease usually heals before a systemic disease, characterized by fever and rash, develops. Arthritis involving multiple joints is commonly seen in rat-bite disease caused by S. moniliformis but it is rare in spirilar infection. Culture of the synovial fluid exudates is frequently negative, suggesting that, in some cases, involvement of the joint may represent a reactive arthritis.

Pediatric septic arthritis affects a single joint in 95% of cases. Involvement of multiple articulations is noted in half of the cases caused by gonococci, and in 7% of those caused by S. aureus or H. influenzae type b,1 and is especially common among patients with rheumatoid arthritis.31 Septic arthritis usually affects the large weight-bearing joints of the lower extremities. In a large series of 781 septic joints diagnosed in 725 patients in two medical centers in Dallas, the knee was affected in 40% of the cases, followed by the hip in 225, the elbow in 14%, the ankle in 13%, the shoulder in 5%, and the wrist in 4%. The sacroiliac, interphalangeal, metacarpal, metatarsal, acromio- and sternoclavicular joints represented less than 1% each. Involvement of the small joints of the hand and feet are overrepresented in K. kingae infections, and the sacroiliac joints are frequently affected in brucellosis. Pseudomonas aeruginosa often causes sternoclavicular joint infection in intravenous drug abusers and is a rare complication of subclavian vein catheterization. Infection of the sternoclavicular joint is frequently associated with adjacent osteomyelitis and may extend posteriorly into the mediastinum.

Most children with septic arthritis present with fever and inflammatory changes over the affected joint. Newborns, as well as young patients infected with low-grade virulence pathogens such as brucellae or K. kingae, may be afebrile at the time of diagnosis, requiring an increased awareness of the possibility of a joint infection.32 Irritability, pain, abnormal (antalgic) posture, restricted range of motion or refusal to move the affected extremity (“pseudoparalysis”), and limping are frequent complaints. Arthritis of the hip is frequently difficult to localize and patients may present with pain referred to the knee or anterior thigh.1 An infected hip is often held flexed, externally rotated, and abducted to relieve intracapsular pressure. Associated osteomyelitis should be suspected if symptoms have been present for several days and the child now has acute worsening of pain suggesting extension of the infection from bone to joint. Although, in most pediatric cases, the source of infection remains occult, associated extra-articular symptoms and signs may provide a clue to the likely bacterial etiology as shown in Table 48–3. Obviously patients should be also carefully evaluated for clinical findings that may suggest diagnoses other than infection, such as rheumatic disorders or metabolic diseases.

The presenting symptoms of Lyme arthritis and gonococcal arthritis differ from typical septic arthritis. Lyme arthritis is a late manifestation of B. burgdorferi infection. In one-fourth of children, clinical features consistent with Lyme disease such as erythema migrans or cranial nerve palsy precede the arthritis. Approximately 90% of cases involved the knee.33 The affected joint is warm and swollen but only mildly tender. Signs and symptoms of systemic illness are uncommon in older children although younger children may have fever at arthritis onset.34 Gonococcal arthritis is often preceded by disseminated gonococcal infection. This syndrome consists of fever, chills, tenosynovitis, polyarthralgias or polyarthritis, and dermatitis. The rash characteristically includes hemorrhagic papules and pustules located on the extensor surfaces of the extremities and over the affected joint. Most patients, as mentioned earlier, have asymptomatic genital, anal, or pharyngeal gonococcal infections. Some patients present without the preceding arthritis-dermatitis, making the condition difficult to distinguish from typical bacterial septic arthritis.

The initial laboratory and radiologic evaluation of a patient with suspected infectious arthritis is summarized in Table 48–4.

Synovial Fluid

The diagnosis of bacterial arthritis in children requires a high index of clinical suspicion. Immediate synovial fluid aspiration followed by microbiological, biochemical, and cytological studies are mandatory.35 If synovial fluid cannot be obtained by needle aspiration at the bedside, the joint should be aspirated with imaging guidance, particularly for joints that are not easily accessible such as the hip, shoulder, or sacroiliac joints.36 Any purulent joint effusion in children should be considered infected until proven otherwise. Although acute rheumatic fever, Reiter's disease, and rheumatoid arthritis can cause a markedly inflammatory synovial fluid, the highest leukocyte counts are seen in the joint fluid of patients with septic arthritis. In children with septic arthritis, there are typically 50,000–200,000 cells per mm3 in the synovial fluid; more than 90% are polymorphonuclear leukocytes (Table 48–5). A synovial fluid white blood cell (WBC) count higher than 50,000 leukocytes per mm3 has been frequently proposed as a cutoff to differentiate between septic arthritis and of noninfectious joint exudates, yet lower counts may be seen in infections caused by gram-negative organisms such as N. gonorrhoeae, K. kingae, and brucellae, and early in the course of bacterial arthritis of any etiology.35

A low synovial fluid glucose concentration (<30–40 mg/dL) suggests infection but the sensitivity of this criterion is only 50%. Low glucose levels can also occur in patients with rheumatoid arthritis. Measurements of protein and lactate content in the synovial fluid aspirate are neither sensitive nor specificity for bacterial arthritis.37 The only definitive proof of an infectious etiology of the joint inflammation is the demonstration of bacteria in the Gram stain or the recovery of an irrefutable pathogen in the culture. A Gram stain should be prepared from a centrifuged specimen (when the fluid volume is sufficient) and carefully examined. Gram's stain smears are positive in 75% of patients with staphylococcal arthritis but in less than half of those infected with gram-negative bacteria,35,38 probably because of a lower bacterial load and the difficulties in recognizing the presence of organisms against the pink-stained fibrin background. The fluid should be inoculated bedside or promptly transported to the microbiology laboratory, seeded onto appropriate media (including a chocolate-agar plate for the isolation of H. influenzae), and incubated in a CO2-enriched atmosphere.

Inoculation of a pediatric blood-culture vial, and preferably one containing antibiotic-binding resins such as the BACTEC™ 9240 Peds Plus bottle39 or the BacT/Alert PedibacT vial (Organon Teknika Corporation, Durham, NC),)40 is also recommended because this method significantly improves the recovery of fastidious organisms and the recovery of organisms in patients already receiving antimicrobial therapy. Anaerobic cultures are not routinely indicated in children in the absence of specific risk factors such as penetrating wounds, human and animal bites, or a sensory neuropathy. Aspiration of an amount of fluid insufficient for a comprehensive laboratory testing is a common event in young children or when a small joint is drained. In this case, performance of a Gram stain and inoculation of a blood-culture vial are probably the best diagnostic options.

Blood

Blood cultures should be drawn in all patients with suspected suppurative arthritis, not only because the accessibility of the specimen compared with drawing a joint exudate specimen, but also because the etiologic agent may be recovered from the bloodstream in up to 50% of cases, even when cultures of the synovial fluid are sterile.35 Genital cultures should be obtained in sexually active adolescents presenting with arthritis and signs compatible with a disseminated gonococcal infection.

Acute-phase reactants such as WBC counts, erythrocyte sedimentation rate and C-reactive protein (CRP) levels are usually elevated in children with septic arthritis. A normal WBC count may be seen in neonates with infected joints. Children with septic arthritis who have a normal CRP at presentation, usually have an elevated value 8–12 hours later. Sequential measurement of CRP levels may be used to follow the disease response to therapy; the CRP is more sensitive than the ESR for this purpose. A favorable clinical course is characterized by an average normalization of CRP values (<20 mg/L) within 10 days, whereas increasing levels may represent therapeutic failure and help in the early recognition of complications.41

Other Studies

Lyme arthritis is suggested by serologic evidence of Lyme disease confirmed by Western blotting in a patient with documented arthritis. By the time symptoms of arthritis develop, virtually all patients have a positive serum IgG test. Since arthritis develops late in the course of infection, IgM response may no longer be present. B. burgdorferi DNA can be detected by PCR in the synovial tissue or joint fluid of most patients. In one study of patients with Lyme arthritis, B. burgdorferi PCR testing of synovial fluid was positive in 70 (96%) of 73 patients with Lyme arthritis who had received no antibiotics or short-course antibiotic therapy.42 In contrast, the test was not positive in any of the 69 control patients diagnosed with other forms of arthritis (e.g., rheumatoid arthritis, degenerative joint disease, osteoarthritis).42

Although only 25–50% of patients with gonococcal arthritis have positive joint fluid cultures, PCR-based assays are extremely sensitive in detecting gonococcal DNA from synovial fluid. Cultures from mucosal surfaces (cervix, urethra, rectum, vagina, or throat) are often positive for N. gonorrhoeae when inoculated onto Thayer-Martin agar and incubated in an enriched CO2 environment within 15 minutes of specimen collection. N. gonorrhoeae may also be detected by ligase chain reaction on first-voided urine specimens and urethral and cervicovaginal swab samples.

Imaging studies are not diagnostic for septic arthritis but are helpful in supporting a clinical suspicion of the disease, detecting concomitant osteomyelitis, and excluding other conditions. The initial radiographic examination is usually normal or may reveal subtle changes such as soft tissue swelling, widening of the joint space, and osteolytic changes suggesting contiguous osteomyelitis. When a suppurative infection of the hip is suspected, roentgenograms should be obtained in both, the “frog-leg position,” as well as with the legs extended at the knees and slightly internally rotated. Displacement of the femoral head laterally and upward and of the obturador internis muscle medially by a distended joint capsule would support the diagnosis.1 Radionuclide imaging may be useful to localize a deep infection in the hip, vertebrae, or the sacroiliac joints. Technetium phosphate joint scintigraphy is positive in any synovitis, whereas sequential gallium imaging, although not specific, may differentiate infection from other causes of joint inflammation.43 Sonography may detect accumulation of intra-articular fluid, and bone scans may be used to localize the joint affected when in doubt. Although magnetic resonance imaging has a high resolution for skeletal system pathology, its use as a routine diagnostic test for pediatric septic arthritis is not currently recommended.

Septic Arthritis in the HIV-Positive Patient

Despite the profound immunosuppressive effect of the human immunodeficiency virus it remains unclear whether HIV-infected individuals have a true excess of skeletal system infections because of the confounding factors of intravenous drug abuse and frequent hospitalizations. Staphylococcus aureus is the most common cause of joint infection among AIDS patients and salmonellae appear to be over-represented in this population. In patients with advanced disease, and particularly in those with a CD4 count <200/mm3, pneumococci, and opportunistic mycobacteria, Nocardia asteroides, Candida species and other yeasts, as well as a variety of fungi have also been reported.44

Culture-Negative Septic Arthritis

Although the accepted proportion of cases of suspected joint infections with negative cultures averages 33%,1 percentages ranging between 16%45 and 60%46 have been reported. These discrepancies may be ascribable to differences in recruitment, inclusion and exclusion criteria, and performance of culture methods.47 The epidemiological profile and clinical presentation of children with failed cultures is frequently similar to that of those in whom a pathogen is recovered, although a trend toward lower body temperature and CRP values on admission, a milder clinical course, and better prognosis have been reported in two patients’ series.48,49 Whether these observations represent infection with fastidious pathogens of lower virulence is unknown. Because of the potential serious consequences of delayed or inadequate antimicrobial therapy, patients with suspected septic arthritis and a negative culture should be administered a full course treatment with a broad-spectrum antibiotic. It is to be expected that in the future, improved microbiological culture methods and use of nucleic acid amplification essays will reduce or eliminate these cases altogether.

Neonatal Septic Arthritis

Invasion of the joint space in neonates may occur during a bacteremic episode, as a result of dissemination of the infection from a contiguous focus of osteomyelitis or, more rarely, by direct inoculation of skin organisms during a femoral venipuncture.50 The source of the preceding bacteremic episode may be the nosocomial transmission of virulent S. aureus organisms, the newborn's normal skin flora, as in coagulase-negative staphylococcal or Candida species infections in premature babies with indwelling intravenous catheters, or acquisition of maternal flora in newborns delivered through a birth canal colonized with Enterobacteriaceae, S. agalactiae or N. gonorrhoeae.1,51 Limited use of an extremity or pseudoparalysis resembling Erb's palsy, inflammatory signs over the affected joint, and multiple joint involvement may be found. A complete sepsis workout, including the performance of a lumbar puncture, is indicated.

Prosthetic Joint Infections

Infections of implanted joint prostheses are rare, and occur in 1–2% of knee arthroplasties, in 0.3–1.3% of hip replacements, and in less than 1% of shoulder arthroplasties.52 The risk for infection is increased after revision procedures and in patients with rheumatoid arthritis. Microbial colonization of the implant may occur at the time of surgery, as a result of spreading from a contiguous wound infection, or from hematogenous seeding. Acute symptoms and signs suggestive of infection may develop in the early post operative period or infection may present as a smoldering painful process with prosthesis loosening several months or years after surgery. Virulent organisms such as S. aureus are recovered in only 20–25% of cases, whereas low-grade pathogens such as coagulase-negative staphylococci, non hemolytic streptococci or Enterococcus species, and anaerobes are isolated in more than half of the cases. Polymicrobial infections are detected in 12–19% of patients. The sensitivity of clinical examination, laboratory, radiographic or nuclear scans is insufficient to diagnose infection in a prosthetic joint, and ultimately, only identification of the etiology in the Gram stain and culture of the synovial fluid aspirate and/or histopathologic evaluation of the periprosthetic tissue confirm the clinical suspicion. The implications of septic arthritis in a prosthetic joint in terms of need for reoperation, potential loss of the prosthesis, and economic burden are substantial. Treatment of infected joint prostheses is challenging, requiring multiple surgical procedures and prolonged antimicrobial therapy and a close collaboration between orthopedic surgeons and infectious diseases specialists.52

Antibiotic Therapy

For the reason that the potential risk for long-term orthopedic sequelae caused by the rapid joint destruction, septic arthritis should be considered a true pediatric emergency requiring a high index of clinical suspicion, early joint drainage, and prompt administration of adequate antimicrobial therapy. The initial antibiotics therapy should be administered through the parenteral route. The penetration of most antibiotics into inflamed joint effusions is sufficient providing that an adequate serum level is achieved. Therefore, antibiotics do not need to be injected intra-articularly; furthermore, this is contraindicated because of the possibility of inducing chemical synovitis.35

The choice of the initial antibiotic therapy should be guided by the results of the Gram stain examination of the fluid, age of the patient, clinical picture, presence of specific risk factors (such as immunodeficiency), potential exposures to organisms such as B. burgdorferi or brucellae, and the local prevalence of antibiotic resistance in organisms such as S. aureus. Once the pathogen is identified and antibiotic susceptibility is determined, antibiotic therapy should be adjusted accordingly. If no pathogen is isolated but the patient is responding adequately, therapy should be continued with the drug chosen originally. If no improvement is noted the possibility of an uncommon infectious etiology or a non infectious process should be entertained and a more extensive clinical and laboratory evaluation, including reaspiration and, eventually, a synovial biopsy should be considered.1

In the newborn, an antistaphylococcal penicillin, such as oxacillin or nafcillin (150–200 mg/kg per 24 hours divided q6h), in combination with a third generation cephalosporin, such as cefotaxime (150–200 mg/kg per 24 hours divided q8h), provide adequate empiric antimicrobial coverage (pending culture and antibiotic susceptibility results). In premature babies with long-term indwelling intravenous catheters, vancomycin (10–15 mg/kg per 24 hours divided in 1–3 doses, according to weight and age) instead of the penicillinase-resistant penicillins is recommended to cover nosocomial coagulase-negative staphylococci.

In preschool age children up to 5 years of age, options include cefuroxime (200–300 mg/kg per 24 hours divided q8h), cefotaxime, or ceftriaxone because each adequately covers gram-positive pathogens such as methicillin-susceptible S. aureus, β-hemolytic streptococci, pneumococci, as well as gram-negative organisms such as H. influenzae type b, K. kingae, N. meningitidis, and most of the Enterobacteriaceae. In children older than 5 years of age, gram-positive bacteria constitute the vast majority of isolates. Anti staphylococcal antibiotics alone would provide adequate coverage pending culture results, unless N. gonorrhoeae is suspected, in which case, addition of ceftriaxone is recommended. In areas where CA-MRSA organisms are prevalent, vancomycin (40–60 mg/kg per 24 hours divided q6h) or clindamycin (30–40 mg/kg per 24 hours divided tid-qid) should be administered, although resistance to the latter has been detected in a few isolates.12 Experience with the use of linezolid for CA-MRSA is limited but the drug appears to be effective for the treatment of suppurative arthritis.53 The use of quinopristin/dalfopristin and daptomycin as alternative drugs for skeletal infections caused by CA-MRSA has not been adequately evaluated in children, and emergence of resistance to the latter drug in the course of therapy for osteomyelitis has been reported.54

Immunocompromized hosts should be given broad-spectrum antibiotics to cover a large variety of common as well as rarer opportunistic organisms. Combinations of vancomycin with ceftazidime (150–200 mg/kg per 24 hours divided q4–6h) or combinations of piperacillin-clavulanate (300–400 mg/kg per 24 hours divided q6–8h) or ticarcillin-clavulanate (200–300 mg/kg per 24 hours divided q4–6h) with an aminoglycoside are currently recommended.1 When joint infections are caused by direct penetrating wounds or bites, anti anaerobic antibiotic coverage with clindamycin should be added. S. aureus arthritis and infections caused by Enterobacteriaceae require 4 weeks of therapy, whereas arthritis caused by other organisms may be adequately treated for a total 2 weeks.

Children with Lyme arthritis may receive oral medications such as amoxicillin, doxycycline (if older than 8 years of age), or cefuroxime initially. If symptoms fail to improve substantially and the diagnosis is certain, ceftriaxone may be used. In children aged 8 years or older, brucellar arthritis is best treated with a combination of oral doxycycline (2–4 mg/kg per 24 hours, up to a maximum of 200 mg/d, divided q12h), and rifampin (15–20 mg/kg/day, maximum 600–900 mg/d, in 1 or 2 divided doses) for 6 weeks. In children younger than 8 years, oral trimethoprim-sulfamethoxazole (10 mg/kg per 24 hours of trimethoprim hours, maximum 480 mg/d, divided q12h and sulfamethoxazole, 50 mg/kg/d, maximum 2.4 g/d) should be used instead of doxycycline.55

Once the patient's condition has improved and surgical procedures are no longer necessary (usually within 1 week), the initial parenteral antimicrobial therapy may be shifted to oral drugs for infections involving the knee, ankle, and other small joints. The synovial fluid antibiotic concentrations after oral administration routinely exceed the serum concentrations by 60% or more.56 Prerequisites for oral therapy include control of infection and inflammation and compliance with planned therapy and monitoring. For the oral regimen of β-lactam antibiotics, a dosage 2–3 times higher than that used for more benign infections is needed.1 In the past, it was recommended to assess the adequacy of the antibiotic dosage and absorption of the drug in patients with osteomyelitis and septic arthritis treated with sequential parenteral-oral antibiotic therapy. The peak bactericidal activity at the steady state (after the third dose of oral antibiotics) was measured in the serum and finding of a level of 1:8 or greater was considered adequate.1 A recent study has suggested that determining antibiotic levels may be unnecessary and patients with staphylococcal osteomyelitis may be safety managed with judicious clinical evaluation and serial CRP determinations.57 Further experience with this simplified treatment approach and with its use in children with septic arthritis or infections caused by other organisms is required before it can be universally recommended.

The appropriate duration of antimicrobial therapy for septic arthritis is not clear but should be individualized based on the organism isolated and the clinical and laboratory response. Minimum criteria for discontinuing antibiotic therapy include resolution of signs and symptoms of infection and normalization of the CRP level. The CRP peaks on the second day of therapy and is normal within 7–9 days. In contrast, the ESR rises slowly over several days, peaks in the first week, and then normalizes slowly during the next 3–4 weeks. Since the CRP increases and decreases much more quickly than ESR, measuring the CRP may be more useful in determining response to therapy. CRP values that remain high or increase again during therapy require careful investigation. Waiting for normalization of the ESR may be an overly conservative therapeutic end point. Most infections require 2 weeks of therapy but septic arthritis caused by S. aureus or gram negative organisms require at least 3 weeks of therapy. Children with associated osteomyelitis should be treated for 4 weeks since a shorter duration of therapy may be inadequate for the treatment of osteomyelitis.

Lyme arthritis is treated with a 4-week course of oral amoxicillin or doxycycline, depending on age. Fewer than 10% of children fail an oral regimen but children with incomplete response to oral antibiotic therapy require intravenous ceftriaxone or penicillin for 14–21 days. Surgical irrigation of the joint is not usually required. Up to 50% of children with Lyme arthritis will have a recurrent episode of arthritis within 6 months of initial treatment.33 Recurrences can be treated with non-steroidal anti-inflammatory agents. The frequency and duration of recurrences, which occur more commonly in females and in older children,34 diminish over time. Patients who receive intra-articular steroids prior to antibiotic treatment for Lyme arthritis appear to have a longer time to resolution of the arthritis.34 Repeating Lyme serologic tests during recurrences does not alter management. Chronic arthritis rarely develops in children with Lyme arthritis.

Gonococcal arthritis should be treated with intravenous ceftriaxone or other broad-spectrum cephalosporins. Skin lesions may continue to develop during the first two days of therapy. Treatment may be switched to oral fluoroquinolones or, if penicillin-susceptible organisms are isolated, amoxicillin to complete 7–10 days of total therapy. Surgical irrigation of the joint is not routinely required. Sexually active adolescents should be evaluated for other sexually transmitted diseases, including C. trachomatis. Prepubertal children require evaluation for sexual abuse.

Anti-Inflammatory Therapy

Results of a recent double-blind placebo-controlled study conducted among 123 children with joint infections in Costa Rica have shown that adding intravenous dexamethasone (0.6 mg/kg per 24 hours divided q8h) for 4 days to appropriate antibiotic and surgical therapy significantly reduced the duration of symptoms and residual dysfunction compared to placebo, supporting the role played by the host's immune response in the degradation of the joint cartilage.58 These promising results need to be confirmed before the administration of corticosteroids can be routinely recommended in patients with septic arthritis.

Surgical Treatment

Because of the deleterious effect of infected synovial fluid on the cartilage layer, removal of the accumulated exudate is recommended. A seminal study conducted by Nelson and Koontz59 in 1966 concluded that needle aspirations (single or multiple, as clinically required) were preferable over open drainage as the initial management of infected joints, and these results have been later confirmed by others. The hip and shoulder are the exceptions to this rule because increased intracapsular pressure of undrained fluid may compromise the vascular supply, resulting in necrosis of the femoral or humeral heads. It has been recently suggested that repeated ultrasound-guided aspirations and irrigation may spare the need for surgical drainage,60 but this conservative approach should be confirmed by additional experience. The indications for surgery in the management of pediatric septic arthritis are summarized in Table 48–6.

Long-term follow-up is required to assess the functional results of septic arthritis in children. A variety of functional sequelae such as limping, decrease range of motion, instability or chronic dislocation of the joint and abnormal bone growth have been reported in 10–25% of children with septic arthritis. Risk factors for a poor orthopedic outcome are listed in Table 48–7.37,61