In a child who lives in or has visited an endemic area (primarily the northeastern and mid-Atlantic
states and parts of Wisconsin and Minnesota), Lyme disease is a possible cause of arthritis. Characteristically, one or two large joints suddenly become swollen and warm, but there is often minimal pain or erythema.¹⁸ The knee is the most frequently affected joint. There may or may not be a history of a tick bite or an erythema migrans rash. However, by the time arthritis develops (usually several weeks after infection), a specific IgG and IgM antibody response in the serum to Borrelia burgdorferi is nearly always present.¹⁹ Synovial fluid leukocyte counts are typically less than 50,000 per mcL but range from 500–100,000 per mcL, usually with a neutrophil predominance.²⁰ Using PCR, B. burgdorferi DNA can be detected in the synovial fluid of patients with Lyme arthritis.²¹ In the small percentage of patients who develop chronic arthritis despite antimicrobial therapy, PCR testing is negative.²¹ Thus, in addition to causing infectious arthritis, B. burgdorferi can apparently cause a form of reactive arthritis in some genetically-predisposed people.²² Chronic arthritis in children with Lyme disease is uncommon; greater than 95% of children treated with an appropriate course of antibiotics have no long-term symptoms.²³

It is not uncommon for a parent to request testing for Lyme disease in a child with nonspecific symptoms, such as headache, fatigue, and arthralgia. In the absence of a history of erythema migrans or objective findings (such as arthritis, carditis, meningitis, or neuritis), the likelihood of Lyme disease is remote. In such children the positive predictive value of serologic testing is less than 5%; that is, nearly all positive tests are false-positives.²⁴ For persons not residing in a Lyme-endemic region, the positive predictive value of serologic tests approaches zero. Testing for Lyme disease in such patients is counterproductive, because a positive test is likely to result in delayed treatment of the correct diagnosis, such as fibromyalgia or chronic fatigue syndrome.²⁵

The most important consideration in suspected septic arthritis of the hip is osteomyelitis of the femoral neck or, less commonly, of a pelvic bone (see the section on osteomyelitis). It is also important to remember that osteomyelitis and septic arthritis sometimes occur together.³⁰

Sometimes called a “sympathetic effusion,” joint fluid near an osteomyelitis closely resembles septic arthritis clinically. However, aspiration of the joint reveals relatively clear fluid, without bacteria present on the Gram stain. Only joint aspiration can distinguish between a purulent and a sympathetic effusion in a joint adjacent to an area of osteomyelitis, although range of motion is usually less limited in a sympathetic effusion. The pus formed in the metaphysis in osteomyelitis may break through into the joint space, especially in joints where the metaphyseal spongiosa lies within the attachments of the joint capsule, as in the hip and shoulder.⁵⁸ To avoid the damaging effects of pressure and pus, it is essential to aspirate such joints to recognize a purulent arthritis. When the joint tap is nonpurulent in suspected septic arthritis of the hip, a bone scan should be done to look for femoral neck osteomyelitis.

Transient synovitis, also called “toxic synovitis” or “irritable hip,” is manifested by unilateral hip pain,
refusal to walk, or limping, and is a frequent cause of arthritis in children. In fact, it is the most common cause of atraumatic limp in children presenting to the emergency department, accounting for about 40% of such visits in one study.⁵⁹ Fever, if present, is usually low-grade. Transient synovitis is much more frequent than septic arthritis of the hip. Nevertheless, the physician must not assume that a child with a swollen, painful hip joint has transient synovitis rather than septic arthritis. The risks of diagnostic joint aspiration performed by a physician who is experienced in this procedure are small compared with the dangers of a destructive septic arthritis. Ultrasound is not helpful in distinguishing transient synovitis from septic arthritis because an effusion is found in both conditions.

Several algorithms have been developed to assist in determining the likelihood of septic arthritis vs. transient synovitis. Kocher et al.⁶⁰ reviewed the charts of all children presenting to Children’s Hospital in Boston with an irritable hip over an 18-year period. Of 262 children, 82 were diagnosed with confirmed or presumed septic arthritis, 86 were diagnosed with transient synovitis, and 114 were excluded. Lack of synovial fluid studies was the most common reason for exclusion, and, presumably, many of these children had transient synovitis. In multivariate analysis, four variables were found to be independent predictors of septic arthritis: a history of subjective fever, non-weight-bearing, ESR 40 mm/hour or greater, and peripheral WBC greater than 12,000 per mcL. The probability of septic arthritis could be calculated based on the number of these variables present (Table 16-4). Using this algorithm, children with three or four predictors should usually undergo hip aspiration in the operating room, given the high likelihood that arthrotomy and drainage will be needed. Children with two predictors may be good candidates for aspiration in the outpatient setting. Depending on clinical suspicion, some children with fewer than two predictors will be appropriate candidates for careful observation without aspiration. Other similar algorithms have been developed.^61,62 It is important to remember that the utility of such algorithms may be lower in children less than 18–24 months of age, in whom transient synovitis is less common and in whom the presentation of septic arthritis is often subtle.

The cause of transient synovitis of the hip is unknown. Symptoms usually last only a few days, although it can be recurrent. Legg-Calve-Perthes disease (aseptic necrosis of the femoral capital epiphysis) in its early stage may cause synovitis that is indistinguishable from transient synovitis, so that patients with synovitis should have careful evaluation and follow-up, with referral to an orthopedic surgeon if symptoms persist.

Cellulitis near a joint may resemble acute arthritis. However, with septic arthritis the erythema is usually symmetrically distributed around the joint, and the borders of redness are diffuse. Motion of the joint is usually exquisitely painful. Cellulitis will typically have more well-demarcated erythema, often with evidence of a local wound, and joint motion causes little discomfort. Despite these generalizations, it is often difficult to tell whether the
joint is involved. Diagnostic aspiration (avoiding the cellulitis, if possible) should usually be done if there is reasonable suspicion of increased fluid in the joint. The risk of introducing bacteria from cellulitis into the joint is small compared with the risk of delay in diagnosing septic arthritis. In certain cases, ultrasound may be performed first and, if there is no joint effusion, arthrocentesis can be avoided. However, the ultrasound must be done in a timely fashion by a radiologist expert in interpreting the results. Sonography is useful for excluding joint effusion in the hip but may be less sensitive in detecting fluid in smaller joints. If no joint fluid is present by clinical or ultrasound examination (and osteomyelitis is not suspected or has been ruled out), blood cultures should be obtained and the patient should be treated with intravenous antibiotics such as oxacillin or cefazolin to cover S. aureus and Group A streptococcus. The patient usually responds promptly when cellulitis without underlying septic arthritis is present, as discussed in Chapter 17.

Bacterial infection of a bursa is usually secondary to a local laceration or abrasion of the skin.⁶³ In children, this condition is most frequently observed in the prepatellar bursa after an abrasion of the knee. Septic prepatellar bursitis can usually be distinguished from septic arthritis of the knee by physical examination, because there is less pain on motion in bursitis. Treatment with drainage and antibiotics is similar to that for a skin abscess.

Reactive arthritis is defined as a nonpurulent arthritis that occurs simultaneously or subsequent to an infection elsewhere in the body. The larger joints are more commonly affected. Onset is typically acute and fever may be present. The most well recognized associations are with various causes of bacterial enteritis and genital infection (see Box 16-1). There remains considerable debate as to whether such arthritis is caused by occult infection in the joint or by an immune response to infection.⁶⁴ It is possible that different mechanisms are operative depending on the pathogen responsible. Nucleic acid of Chlamydia trachomatis and, less frequently, C. pneumoniae has been found in the synovial tissue of patients with reactive arthritis.^65,66 Chlamydia DNA is present in the synovial tissue of a small percentage of asymptomatic subjects as well.⁶⁷

Some patients are genetically predisposed—there is a higher incidence of the HLA-B27 phenotype in patients with reactive arthritis—but this does not explain all cases. Some infectious agents, such as meningococcus⁶⁸ and Group A streptococcus,^69,70 can cause either purulent arthritis or reactive arthritis. The mechanism of arthritis for many viral infections may be immune-related, and these can be considered a form of reactive arthritis. Several vaccines have been anecdotally associated with the development of reactive arthritis.^{71,72,73,74,75,76,77} “Reiter’s syndrome” is the term used for the subset of patients with reactive arthritis who also have nongonococcal urethritis and ocular inflammation.⁷⁸

Many systemic inflammatory diseases may have arthritis as part of the clinical presentation. These include collagen vascular diseases such as JRA,⁷⁹ systemic
lupus erythematosis,⁸⁰ mixed connective tissue disease,⁸¹ juvenile ankylosing spondylitis,^81a and polymyositis and dermatomyositis.⁸² Other systemic diseases that may present with arthritis include acute rheumatic fever,⁸³ inflammatory bowel disease,⁸⁴ Kawasaki disease,⁸⁵ sarcoidosis,^85a and Henoch-Schöenlein purpura.⁸⁶ Serum sickness is also a consideration, particularly in the child who has received medication in the previous 2 weeks.⁸⁷ Approximately 5–10% of children with cystic fibrosis have episodic arthritis that is probably immune-mediated.⁸⁸

Several malignant conditions in childhood may initially manifest joint involvement, especially leukemia and neuroblastoma.⁸⁹ Laboratory clues to the presence of malignancy include anemia, elevated erythrocyte sedimentation rate (ESR) in the presence of a normal or decreased platelet count, and elevated serum lactate dehydrogenase (LDH).⁹⁰

Trauma with bleeding into the joint may mimic septic arthritis. A history of only mild trauma resulting in significant joint swelling should raise suspicion for child abuse or an underlying coagulation disorder. It should also be kept in mind that a history of recent trauma to the joint is very common in children with septic arthritis, and such a history should not decrease suspicion for this diagnosis.

Penetrating foreign body injury to the joint may result in a presentation identical to septic arthritis.^91,92 Cultures are usually sterile, but such injury can occasionally inoculate bacteria into the joint resulting in a traumatic septic arthritis.^93,94 Clenched fist human bite injuries commonly cause infection of the underlying metacarpal-phalangeal joint.⁹⁵ Early surgical exploration is indicated for such injuries, and prophylaxis with amoxicillin-clavulanate is appropriate to cover the commonly implicated organisms (streptococci, S. aureus, Eikenella corrodens, and anaerobes).

Radiologic examination may be useful to indicate the presence of fluid in the hip joint, which is difficult to detect by physical examination (Figs. 16-1 and 16-2), although ultrasound is superior for this purpose. Plain film x-rays are most helpful to detect an unsuspected fracture or chronic bone or joint disease, and thus their use is justified. However, they are not helpful in excluding the diagnosis of septic arthritis.

Ultrasound can reliably detect fluid in the hip joint,⁹⁶ and may be useful in detecting fluid in other joints as well.⁹⁷ However, sonography cannot distinguish whether a joint effusion is purulent, sterile, or hemorrhagic.⁹⁶

Scanning with radioactive technetium (^99mTc) is especially useful to detect septic arthritis in joints that are difficult to examine clinically, such as the sacroiliac area.⁹⁸ On bone scan, septic arthritis usually
reveals diffuse, faintly increased tracer uptake on both sides of the joint. In osteomyelitis, the increased tracer uptake is usually unilateral and more intense, but differentiating the two conditions may be difficult.⁹⁹ In some cases of septic arthritis of the hip, fluid in the joint under pressure causes impaired perfusion and decreased tracer uptake (“cold-hip” sign).¹⁰⁰

These studies are of limited usefulness, because joint fluid and synovial enhancement will be seen on CT or MRI with arthritis of any cause.⁹⁹ Their main utility in this setting is in looking for an associated osteomyelitis.

If a joint effusion is suspected because of clinical or imaging findings, an arthrocentesis should be done. Some of the fluid should be put in a tube containing an anticoagulant for use in studies that cannot be done with clotted blood. Such clotting is due to fibrin, which is not present in normal joints. A white cell count and differential should be done, and the joint fluid should be Gram stained, noting the frequency of leukocytes as well as looking for bacteria. Bacterial antigen detection tests are neither approved for use with joint fluid nor helpful in diagnosing or excluding septic arthritis.¹⁰¹

The highest yield for bacterial culture is obtained by inoculating several milliliters of joint fluid directly into a standard blood culture bottle for use in an automated detection system.¹⁰² A small amount of fluid can also be inoculated into a lysis-centrifugation tube, but these tubes are not as commonly available, and whether their use increases the yield over that of the standard blood culture bottle is unclear.¹⁰³ Either of these methods is superior to directly plating synovial fluid onto solid media, especially for recovery of fastidious organisms such as Kingella kingae.¹⁰⁴ The microbiology laboratory should be notified if an unusual organism is suspected because of the patient’s exposure history. Stain and culture for fungi and mycobacteria are not necessary for most cases with an acute presentation. These tests should be considered in the immunocompromised host, the patient with a subacute or chronic course, or if an initial joint tap was culture negative and the response to empiric antibacterial therapy has been slow. Keep in mind, however, that the most common reason for a poor response to therapy is inadequate surgical drainage. Examination for crystals, an important part of synovial fluid analysis in adults, is rarely necessary in children.

A presumptive diagnosis of septic arthritis can be made by immediate examination of the joint fluid, which is usually cloudy. Bacteria and a predominance of segmented neutrophils are often seen on the Gram stain. Leukocyte counts greater than 50,000 per mcL are strongly suggestive of bacterial infection, although some patients will have lower counts. Occasionally, patients with other causes of arthritis will have leukocyte counts greater than 50,000 per mcL.¹⁰⁵ Decreased glucose and increased
protein concentrations in the joint fluid are supportive but not universal findings and should not be relied upon. Only about 30–60% of synovial fluid cultures will be positive, even in the child who has not received recent antibiotics.^{4,106,107,108} Thus, a negative culture does not rule out the possibility of bacterial arthritis.

Drainage of pus from a joint may be done by intermittent aspiration, by open incision with drainage (followed by continuous suction drainage) or by arthroscopy. The risk of severe deformity from septic arthritis in children leads most orthopedic surgeons to argue for open decompression and removal of pus. However, repeated aspiration may be acceptable for the knee, elbow, or ankle, if there is rapid improvement. In the hip and shoulder, open incision and drainage is advisable.

If the joint fluid shows more than 10,000 leukocytes per mcL with a predominance of neutrophils, the working diagnosis should be “purulent arthritis, probably septic.” Antibiotic therapy should not be delayed until culture results are available. Rather, immediate intravenous therapy should be given based on the Gram stain. If the Gram stain is negative, then antibiotic therapy should be based on the most frequent organisms recovered. In most settings, empiric therapy using an agent with good activity against S. aureus, such as oxacillin or cefazolin, is appropriate. Vancomycin may be used in areas with a high prevalence of community-acquired MRSA. Clindamycin is also an option, depending on local susceptibility profiles.^109a If S. pneumoniae, H. influenzae, Salmonella, or N. gonorrheae are suspected, ceftriaxone or cefotaxime is added. For the neonate, oxacillin plus cefotaxime or gentamicin may be used to cover the most likely pathogens (S. aureus, Group B streptococcus, and enteric gram-negative rods). For all patients, as soon as definite culture and sensitivity results are obtained, the initial therapy can be changed, if necessary. If the isolate (usually S. aureus) is shown to be sensitive to a β-lactam antibiotic, this should be used preferentially because of greater inherent antistaphylococcal activity than vancomycin.

Rarely, slow response to therapy with a β-lactam antibiotic is the result of bacterial tolerance, rendering the antibiotic bacteriostatic instead of bactericidal.¹¹⁰ More commonly, poor response is secondary to inadequate drainage. Culture-negative septic arthritis is common and should be treated as aggressively as culture-positive cases,¹⁰⁷ with therapy that ensures good coverage against S. aureus (the most frequent and serious cause).

Intra-articular antibiotics have not been shown to improve outcome.¹¹¹ In addition, their use may exacerbate the inflammatory response of the synovium. A recent prospective, randomized study of septic arthritis in children reported improved outcome among patients receiving a 4-day course of intravenous dexamethasone compared with placebo controls.^111a However, the incidence of residual joint dysfunction was higher in the control group than is generally seen. These findings require replication in larger studies before the use of carticosteroids in the treatment of septic arthritis becomes routine.

Lyme arthritis is treated effectively with a 4-week course of doxycycline for children older than 8 years and with amoxicillin for those 8 years old or younger.²⁴ For the few patients with persistent arthritis despite appropriate therapy, a single
14–21-day course with ceftriaxone is sometimes used, although this approach has been shown to be of no benefit in controlled trials.¹¹² Prolonged use of intravenous antibiotics in patients with a history of possible Lyme disease and persistent nonspecific symptoms is inappropriate. This practice has been associated with severe complications from the use of central catheters, including death.¹¹³

Initial therapy should be by the intravenous route. It has become common practice to switch to high-dose oral therapy after about a week in the child who has an excellent initial response to therapy, with normalization of temperature and decreased pain and swelling of the joint. Other criteria sometimes used include (1) decreasing ESR or CRP, (2) assurance of compliance, (3) identification of the organism, and (4) documentation of adequate serum levels of antibiotics. Serum bactericidal levels have also been advocated. In this test, the infecting organism must be recovered from the blood or joint fluid. The patient’s serum is obtained at the expected peak or trough concentration of the oral antibiotic and serially diluted. If the organism is killed at a 1:8 dilution at peak or 1:2 dilution at trough, the oral antibiotic is likely to be effective.¹¹⁴ For most antibiotics with time-dependent killing (such as the β-lactam antibiotics), trough titers are expected to be more predictive of therapeutic efficacy, and this was found to be true in a study of adults.¹¹⁵ In practice, serum bactericidal testing is technically difficult, and very few microbiology laboratories offer this service.