Septic Arthritis

Septic arthritis of the hip in a child constitutes a surgical emergency. Prompt recognition, diagnosis, and treatment of this entity are essential. Bacterial infection of the joint space results in an inflammatory effusion consisting of up to 90% polymorphonuclear cells (PMNs).²¹ The release of proteolytic enzymes by PMNs and bacteria, in conjunction with increased intra-articular pressure, can result in rapid and irreversible hyaline cartilage degradation in as little as 6 hours. If unrecognized or left untreated, this process can result in joint destruction with subsequent severe deformity and devastating lifelong disability. Joint infections can ultimately lead to disseminated infection, systemic bacterial sepsis, multiorgan failure, and death. However, if recognized and treated in a timely fashion with surgical drainage and appropriate antimicrobial therapy, a good outcome with minimal sequelae can be expected. Poor outcomes are most closely associated with delay in diagnosis. Other factors related to poor outcome include patient age younger than 6 months, prematurity, Staphylococcus species infection, and concomitant osteomyelitis.

Septic arthritis may occur as a result of hematogenous seeding, local spread from a contiguous infection, or primary seeding of a joint secondary to surgery or trauma. In the hip, shoulder, ankle, and elbow (90% of cases), septic arthritis often develops after metaphyseal osteomyelitis with subperiosteal erosion, abscess formation, and subsequent joint communication (because of the intra-articular location of the metaphyses in these joints). Premature and immunocompromised children are at greater risk. The most commonly identified causative organisms are Staphylococcus aureus, group A streptococci, Streptococcus pneumococcus, other gram-negative organisms (may be seen in special hosts or after trauma-klebsiellae, salmonellae, kingellae), and Neisseria gonorrhea.²² Before the advent of an effective vaccine, Haemophilus influenzae type B was responsible for up to 40% of cases but has largely disappeared because of successful immunization programs.^23,24 In the neonate, group B b-hemolytic streptococcus and gram-negative bacilli are common causative agents in addition to S. aureus. Gram-negative infections are common among intravenous (IV) drug abusers. Mycobacteria and fungi must be considered in patients with chronic infection.

Diagnosis

Children with septic arthritis will often have a history of recent upper respiratory or local soft-tissue infection, as well a recent course of antibiotics. A high level of suspicion is important when evaluating premature and immunocompromised infants. Septic arthritis is more commonly seen in boys and is most common in children younger than 2 years. Patients may report pain, stiffness, and malaise, but such a history is often impossible to ascertain from small children or infants. Fever, limp, inability or refusal to weight bear, limited and painful range of motion (secondary to capsular stretching), erythema, warmth, tenderness, and swelling are common physical findings. The child may complain of anterior hip, groin, thigh, or knee pain. The child may lie with the hip in external rotation, adduction, and mild flexion to maximize joint volume and decrease capsular stretch. Physical examination of the infant can be challenging and may only demonstrate limited or a lack of active motion in the affected extremity (pseudoparalysis).

Laboratory tests that are frequently ordered in cases of suspected septic arthritis of the hip include C-reactive protein (CRP) level, erythrocyte sedimentation rate (ESR), complete blood cell count (CBC) with differential, blood cultures, and throat culture/rapid strep test. Lyme titers are often added in endemic areas. Plain radiographs of the hip or pelvis are routinely ordered. Ultrasound may be useful to detect joint effusion. Magnetic resonance imaging (MRI) may be useful to detect associated osteomyelitis. The definitive diagnosis of septic arthritis is made by joint aspiration and analysis of joint fluid. Joint fluid white blood cell (WBC) count and cultures establish the diagnosis. Organisms can be detected on joint fluid Gram stain or culture. Cases with negative cultures but increased joint fluid WBC counts (>50,000/mm³) are considered to be presumed septic arthritis or culture-negative septic arthritis.

Treatment

Poor outcome is most closely associated with delay in diagnosis with subsequent delay in treatment. Once septic arthritis is diagnosed, prompt treatment with adequate surgical drainage and empiric IV antimicrobial therapy is cardinal. Ideally, synovial and blood cultures are obtained before treatment to increase the likelihood of identifying an organism. Open surgical irrigation and debridement to remove the microorganisms, host and bacterial enzymes, and particulate debris is indicated for most patients. However, arthroscopy has been used successfully in the shoulder, elbow, knee, and ankle, but it has not been widely utilizedin the hip. Empiric IV antimicrobial therapy should be initiated as soon as cultures are obtained. Empiric coverage must include an anti-staphylococcal agent, either a b-lactamase–resistant penicillin or a first-generation cephalosporin. Cefazolin is chosen as the drug for initial empiric therapy because it is effective against S. aureus, group A streptococcus, and S. pneumococcus, which should account for nearly all of the organisms among normal hosts with acute hematogenous osteomyelitis. Gram-negative coverage is indicated in the neonate (causative organisms may also include group B streptococci and gram-negative bacilli) and adolescents (to cover gonococcus). In children who are allergic to penicillin, clindamycin can be used. Ceftriaxone should be considered for coverage of gonococcal or Lyme arthritis. The antimicrobial regimen is adjusted according to culture speciation; if another organism is identified as the causative agent, an infectious disease consultation should be sought. IV antibiotics are continued for 72 hours. If the child shows evidence of clinical improvement (afebrile, decreased localized swelling, decreased/no pain, increased range of motion) and lyme titers are negative, conversion to oral antibiotics can be considered. The criteria for oral antibiotics include diagnosis within 4 days of the onset of symptoms, no concurrent osteomyelitis, and an ability to tolerate and be rigorously compliant with taking oral antibiotics. Patients treated with cefazolin may be given cephalexin 100 mg/kg/day divided in 4 daily doses (maximum dose, 4 g/day), and those treated with ceftriaxone can receive cefixime 8 mg/kg/day every 12 to 24 hours (maximum dose, 400 mg/day) for 21 days. The median duration of IV antibiotics is 5 to 15 days, and the total duration of antibiotics has been 4 to 6 weeks in uncomplicated cases.

Imaging

Ultrasonography has been demonstrated good accuracy in detecting an effusion associated with septic arthritis of the hip in children. Zamzam²⁵ retrospectively studied 81 children with septic arthritis and 73 children with transient synovitis, finding sensitivity of 86.4% and specificity of 89.7%. Dörr and colleagues²⁶ also found ultrasound to be useful in detecting effusion in 204 acutely irritable hips. In addition, they found certain characteristics, such as synovial hypertrophy and capsular thickening, to be helpful in differentiation between septic arthritis and transient synovitis. In a prospective analysis of 500 children with a painful hip or limp, Miralles and researchers²⁷ found ultrasonography to be useful in the detection of effusion. In addition, these authors showed that transient synovitis had resolution of the effusion at repeat ultrasound 2 weeks later.

MRI has been shown to be useful in detecting associated osteomyelitis but may also be useful in differentiating septic arthritis from transient synovitis. In a retrospective case series of 49 children with transient synovitis and 18 children with septic arthritis, Yang and investigators²⁸ found that patients with transient synovitis were more likely to have contralateral effusion and absence of signal intensity in the surrounding soft tissue and bone marrow.

Laboratory Tests

Clinicians have empirically utilized different labora-tory variables to help distinguish between septic arthritis and transient synovitis. Bennett and Namnyak²⁹ emphasize leukocyte count in children older than 3 years. Morrey and coauthors²⁰ and Klein and resear-chers²¹ accent increased ESR. Chen and colleagues³⁰ found both leukocytosis and increased ESR to be common. However, because these retrospective case series of children with septic arthritis did not have a comparison group of children with transient synovitis, they were unable to identify differences between the groups and were unable to determine the diagnostic performance of assorted variables.

On comparing children with septic arthritis and transient synovitis in retrospective case series, poor diagnostic utility has been found for various screening clinical, laboratory, and radiographic data because of substantial overlap between the two groups. Molteni²² retrospectively reviewed 97 patients with so-called nonspecific arthritis and 37 patients with septic arthritis of various joints including the hip, knee, ankle elbow, and wrist. Diagnoses were established presumptively, with joint fluid analysis of 12 of 97 patients with nonspecific arthritis and of 32 of 37 patients with septic arthritis. Fifty percent of the patients with septic arthritis had negative cultures and joint fluid WBC count varied from 5 to 385 3 1000/mm³. Without statistically comparing the two groups, Molteni²² describes the clinical and laboratory findings in each (septic arthritis vs. nonspecific arthritis): age (5.8 vs. 6.0 years), male sex (22/37 vs. 49/97), temperature >101°F (82% vs. 38%), serum WBC count (7.23 vs. 6.3 × 1000/mm³), and ESR (50 vs. 35 mm/hr). Because of the similarities in values and the extent of overlap, he concludes that reliable data were not found to distinguish the two different joint processes. Del Beccaro and colleagues³¹ also found that overlap between screening variables impeded the differentiation of septic arthritis and transient synovitis of the hip in children. They compare 94 children with transient synovitis (defined retrospectively via clinical course) with 38 patients with septic arthritis (defined as pyarthrosis). In the transient synovitis group, 21 of 94 patients had joint fluid analysis, and in the septic arthritis group, 28 of 38 patients had joint fluid cell counts. They found significant (P < 0.05) univariate differences between the two groups with respect to the following factors (septic arthritis vs. transient synovitis): temperature (38.1°C vs. 37.2°C), ESR (44 vs. 19 mm/hr), PMN (7.8 vs. 6.3 × 1000/mm³), and bands (903 vs. 295/mm³). Of note, with the numbers available for study, serum WBC count could not be shown to differ significantly (P > 0.05) (13.2 vs 11.2 × 1000/mm³). Although ESR was significantly (P < 0.05) different between the two groups, the substantial amount of overlap made it a poor discriminator. The range of ESR for the septic hip group was 6 to 90 mm/hr (with 24% of patients with ESR <20 mm/hr), and the range for the synovitis group was 1 to 125 mm/hr (with 28% of patients with ESR >20 mm/hr). By combining ESR >20 mm/hr with temperature >37.5°C, they were able to identify 97% of the septic arthritis cases; however, those parameters also included 47% of the transient synovitis cases. Kunnamo and coworkers³² reviewed 278 children younger than 16 years with various arthritides, including transient synovitis, juvenile arthritis, septic arthritis, serum sickness, enteroarthritis, and Schonlein–Henoch purpura. They found the patients with septic arthritis (n = 18) to differ significantly (P < 0.05) from the other patients with respect to CRP level, temperature >38.5°C, and serum WBC count >12 (×1000/mm³).²⁰ Levine and colleagues³³ emphasize the value of CRP in the diagnosis of septic arthritis of the hip. In a retrospective review of 133 children with an irritable joint, CRP level had better predictive ability than ESR. In particular, if the CRP level was less than 1.0 mg/dL, the probability that the patient did not have septic arthritis was 87%.

Prediction Rules

Prediction rules based on presenting variables have been developed to differentiate septic arthritis from transient synovitis of the hip. Kocher and reviewers³⁴ retrospectively analyzed 82 children with septic arthritis and 86 children with transient synovitis of the hip treated at Children’s Hospital Boston from 1979 to 1996. Diagnoses of true septic arthritis, presumed septic arthritis, and transient synovitis were made based on joint fluid WBC, blood and joint fluid cultures, and clinical course. Using multivariate analysis, they identified four independent predictors of septic arthritis (history of fever, non–weight bearing, ESR ⩾40 mm/hr, and serum WBC >12,000/mm³). The predicted probability of septic arthritis was calculated for all 16 combinations of these 4 predictors (Table 30-1) and was summarized as 0.2% for 0 predictor, 3.0% for 1 predictor, 40.0% for 2 predictors, 93.1% for 3 predictors, and 99.6% for 4 predictors (Table 30-2). This prediction rule demonstrated excellent diagnostic performance with an area under the receiver operating characteristic (ROC) curve of 0.96 (Fig. 30-1).

FIGURE 30-1 Receiver-operating characteristic curves for the clinical prediction rule of Kocher and coworkers34,35 in the original patient population (area under curve = 0.96) and the new patient population (area under curve = 0.86).

(Data from Kocher MS, Zurakowski D, Kasser JR: Differentiating between septic arthritis and transient synovitis of the hip in children: An evidence-based clinical prediction algorithm. J Bone Joint Surg Am 81-A:1662-1670, 1999; and Kocher MS, Mandiga R, Zurakowski D, et al: Validation of a clinical prediction rule for the differentiation of septic arthritis and transient synovitis of the hip in children. J Bone Joint Surg Am 86-A:1629-1635, 2004.)

These authors then validated this prediction rule in a new series of patients.³⁵ They prospectively analyzed 51 children with septic arthritis of the hip and 103 children with transient synovitis of the hip treated at Children’s Hospital Boston from 1997 to 2002. Using the same diagnostic parameters and similar analysis, they identified the same four multivariate predictors of septic arthritis (history of fever, non–weight bearing, ESR ≤ 40mm/hr, and serum WBC 12,000/mm³). The predicted probability of septic arthritis demonstrated similar but not identical values (see Table 30-1) and was summarized as 2.0% for one predictor, 9.5% for one predictor, 35.0% for 2 predictors, 72.8% for 3 predictors, and 93.0% for 4 predictors (see Table 30-2). The diagnostic accuracy showed diminished but still good performance with an area under the ROC curve of 0.86.

Jung and colleagues³⁶ performed a similar study to establish a prediction rule for the differentiation of septic arthritis and transient synovitis of the hip (Fig. 30-2). In a retrospective series of 97 children with transient synovitis of the hip and 27 children with septic arthritis of the hip, they identified five independent multivariate predictors of septic arthritis: temperature >37°C, ESR >20 mm/hr, CRP level >1.0 mg/dL, serum WBC count >11,000/mm³, and >2 mm joint space difference on radiographs. Four of these five predictors were similar to those of Kocher and reviewers.³⁴ They developed an algorithm based on the 32 combinations of the 5 predictors. The area under the ROC curve for their prediction rule was 0.986 (Table 30-3). This prediction rule has not been validated in a new patient population.

FIGURE 30-2 Receiver-operating characteristic curve for clinical prediction rule of Jung and colleagues.³⁶

(Data from Jung ST, Rowe SM, Moon ES, et al: Significance of laboratory and radiologic findings for differentiating between septic arthritis and transient synovitis of the hip. J Pediatr Orthop 23:368-372, 2003.)

Luhmann and coworkers³⁷ developed a new clinical prediction rule and analyzed the performance of the prediction rule of Kocher and reviewers³⁴ in a retrospective series of 47 septic hips and 118 transient synovitis hips at St. Louis Children’s Hospital from 1992 to 2000. Their multivariate model identified three predictors of septic arthritis: history of fever, serum WBC >12,000/mm³, and prior healthcare visit. Two of the four predictors from the algorithm of Kocher and reviewers³⁴ were not predictive in their model: ESR and non–weight bearing. The area under the ROC curve for their model was 77.1%, and the area under the ROC curve for the algorithm of Kocher and reviewers³⁴ in their population was 79.9%.

Caird and researchers³⁸ also developed a prediction rule for the differentiation of children with septic arthritis versus transient synovitis of the hip. In a prospective study of 53 children who underwent aspiration for an acutely irritable hip over a 4-year period at Children’s Hospital of Philadelphia, they identified 5 predictors of septic arthritis: fever >38.5°C, CRP level >2.0 mg/L, ESR >40 mm/hr, refusal to bear weight, and serum WBC >12,000/mm³. The four predictors from the algorithm of Kocher and reviewers³⁴ were predictive in this model, and CRP level showed enhanced diagnostic performance over ESR. The probability of septic arthritis was 16.9% for 0 predictors, 36.7% for 1 predictor, 62.4% for 2 predictors, 82.6% for 3 predictors, 93.1% for 4 predictors, and 97.5% for 5 predictors (Table 30-4).

Clinical Practice Guidelines

Kocher and coworkers³⁹ developed a clinical practice guideline (CPG) for the diagnosis and management of patients with septic arthritis of the hip (Fig. 30-3). The diagnostic arm of the algorithm included laboratory analysis of CRP level, ESR, CBC count with differential, Lyme titer, blood culture, throat culture, and antistreptolysin O titer (ALSO). The imaging arm of the algorithm included radiographs and ultrasound. In a case–control study of the performance of this algorithm, the authors found that CPG patients had significantly greater rates of obtaining initial and follow-up CRP tests (93% vs. 13% and 70% vs. 7%), lower use of initial bone scan (13% vs. 40%), lower rates of presumptive drainage (13% vs. 47%), greater compliance with recommended antibiotic therapy (93% vs. 7%), faster change to oral antibiotics (3.9 vs. 6.9 days), and shorter hospital course (4.8 vs. 8.3 days). No significant differences were found for outcome variables of readmission, recurrent infection, recurrent drainage, development of osteomyelitis, septic osteonecrosis, or limitation of motion. They conclude that patients on the septic arthritis CPG had less variation of process and improved efficiency of care, without a significant difference in outcome.

FIGURE 30-3 Clinical practice guideline for septic arthritis.

(Adapted from Kocher MS, Mandiga R, Murphy J, et al: A clinical practice guideline for septic arthritis in children: Effi cacy on process and outcome for septic arthritis of the hip. J Bone Joint Surg Am 85-A:994-999, 2003, with permission of the Journal of Bone and Joint Surgery.)

TABLE 30-6 Summary of Recommendations