Sections of text from this chapter were written for the Health Technology Assessment (HTA) programme application, but were published separately as Faust SN, Clark J, Pallett A, Clarke NM. Managing bone and joint infection in children. Arch Dis Child 2012;97:545–531 and are reproduced with permission from the British Medical Journal.
Osteomyelitis and septic arthritis in children
Osteomyelitis (OM) is inflammation of the bone accompanied by bone destruction,2 usually due to bacterial infection. It is an acute process, but if not treated effectively the inflammation can become chronic, leading to the development of sequestra and fistulae.3 OM and septic arthritis (SA) can both be divided into three types according to the source of the infection: haematogenous; secondary to contiguous infection; and secondary to direct inoculation. Haematogenous OM can present acutely or subacutely as a more indolent, progressive process, with symptoms present for > 2 weeks.4 In children, OM most often affects the long bones (femur 36%, tibia 33%, humerus 10%, pelvis 2.8%).5 Single-site infection is most common, but 5–20% of children have multifocal OM.6 SA is acute infection of synovial joints,7,8 usually secondary to bacteraemia. The infection affects the synovial membrane and the joint space. In younger children, the capsule of the joint often extends to the metaphysis, and damage to the cortex can lead to SA secondary to OM and vice versa. The epiphyseal growth plate can also be affected, causing growth discrepancies and long-term disability or permanent joint destruction if the acute infection is not treated promptly.3
The estimated incidence for both OM and SA arthritis in Western populations is between 5 and 12 cases per 100,000 children per year.3 Half of children with acute haematogenous OM are aged < 5 years.3,8 Boys are 1.2–3.7 times more likely to be affected by osteoarticular infection (OAI) than girls.3 The incidence in Southampton from 1979 to 1997 was between 1.4 and 10.5 cases per 100,000 per year,9 and in Newcastle from 1991 to 1999 was 7 cases per 100,000 for SA and 11 cases per 100,000 for OM (J Clark, Great North Children’s Hospital, 2011, unpublished data). Recent unpublished national data from England show that the admission rate for OM in children aged 0–18 years has varied between 4.8 and 7.0 per 100,000 child-years (M Sharland, St George’s Hospital London, 2011, personal communication). Subacute OM appears to be have increased in recent years,10 and is reported to be 5 per 100,000 children in Norway.11 Neonatal infection can occur in preterm or term-born babies and is associated with a wider range of causative organisms (see below)12 and potential complications. Neonatal vascular anatomy allows infection within the bone to reach the growth plate or joint in 76% of neonatal osteomyelitis cases.13
The pathogens implicated in paediatric bone and joint infections commonly include meticillin-sensitive Staphylococcus aureus (MSSA) (44–80%)8,14,15 and Kingella kingae (14–50%; higher in children aged < 36 months)8,15–19 and more rarely meticillin-resistant S. aureus (MRSA) (rare in the UK but found in 40–50% of cases in the USA),20,21 Panton–Valentine leukocidin (PVL) MSSA,22,23 group A Streptococcus (GAS), group B Streptococcus (GBS) (neonates),12,24 non-typeable Haemophilus spp. (incidence unknown), Haemophilus influenzae type b (in non-immunised or immunodeficient children), Escherichia coli (neonates),12,24
Streptococcus pneumoniae25 and coagulase-negative Staphylococcus (subacute). Very rarely (mostly in immunocompromised individuals) implicated are Pseudomonas aeruginosa (usually associated with inoculation injuries; therefore, in children aged > 1 year), Neisseria gonorrhoeae, Neisseria meningitidis (neonates, adolescents), Mycobacterium tuberculosis (older children as OAI develops 2 years after primary infection), Salmonella spp. (children with sickle cell disease),26
Bartonella henselae, N. gonorrhoeae, non-tuberculous mycobacteria (associated with defects of the interferon gamma pathway), Klebsiella spp., Bartonella henselae, Fusobacterium (often multifocal), Aspergillus and Candida albicans (neonates, children with damaged bone).
The pathogens most frequently seen according to age are:
neonates: GBS, MSSA, E. coli and other Gram-negative bacteria, and C. albicans
children aged < 2 years: MSSA, K. kingae, S. pneumoniae, H. influenzae type b, non-typeable Haemophilus spp., E. coli and MSSA PVL
children aged 2–5 years: MSSA, K. kingae, GAS, S. pneumoniae, H. influenzae type b, non-typeable Haemophilus spp., Pseudomonas spp., coagulase-negative Staphylococcus (subacute) and MSSA PVL
children aged > 5 years. MSSA and MSSA PVL.
Clinical features
The clinical features of OM and SA are dependent on age, site of infection and type of disease. The diagnosis and management of OAI in children should ideally be multidisciplinary, including paediatricians and orthopaedic surgeons with radiologists and microbiologists. The diagnosis of OM or SA is made on the basis of clinical presentation, laboratory tests, imaging and, where available, microbiology results.1
White blood cell count, C-reactive protein and erythrocyte sedimentation rate
The white blood cell (WBC) count is an unreliable indicator of an OAI as, in many cases, it remains normal throughout the infection.27 The inflammatory markers erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) are more reliable, although normal values also do not exclude OM.28 CRP levels are most sensitive (elevated in up to 98% of cases)7,8 but are not specific for bone or joint infection. Two studies have shown that CRP increases and also decreases faster than ESR, predicting recovery with more sensitivity than the ESR or the WBC count.28,29 Differences in the causative organism may also cause differences in the acute-phase markers. Patients with OM caused by PVL-expressing S. aureus isolates have been found to have significantly higher mean values for ESR at admission, and higher maximum CRP, ESR and absolute neutrophil counts at presentation, than patients whose isolates were PVL negative.23 Other markers remain unproven. In a small study, procalcitonin has not shown benefit over CRP.30
Imaging
Imaging is of great importance in the diagnosis of acute OM. Where available, magnetic resonance imaging (MRI) with enhancement shows the best results regarding sensitivity and specificity of diagnosis of both OM and SA (sensitivity 97% and specificity 92% in SA;31,32 sensitivity 97–100% in OM).7 However, as young children often require an anaesthetic to undergo MRI, and MRI is not immediately available in all UK centres, MRI is not widely used in the UK in the initial diagnosis.
A technetium (99mTc) radionuclide bone scan also has high sensitivity and specificity in the diagnosis of OM,33 but, as a result of the radiation burden, is now used less often except in difficult cases, and is not useful in discitis. In SA, a bone scan may be used to exclude underlying OM following aspiration and commencement of empirical therapy. A bone scan is especially useful where there is a suspicion of multifocal disease, but may give false-negative results in infancy, and sensitivity is reduced for the first 48 hours. New nuclear medicine technologies are available in some centres to combine computerised tomography (CT) with a low-dose radioactive substance, single-photon emission CT, which may be useful in increasing the resolution of nuclear medical images.34
Plain radiography is less helpful than other imaging techniques as osteolytic changes or periosteal elevation occur most often 10–21 days after the onset of symptoms.2,8,35 However, once apparent, the extent of bony change provides a good correlate to the severity of the disease. Plain radiographs also provide a baseline for comparison of subsequent change. Radiographic changes are frequently seen in subacute OM, but can be confused with malignancies such as Ewing’s sarcoma or osteoid osteoma.13 In SA, plain radiographs are of limited use. In discitis, lateral radiographs of the spine 2–3 weeks into the illness often will reveal disc space narrowing with erosion of the vertebral end plates of the contiguous vertebrae. In vertebral OM, radiographs initially show localised rarefaction of a single vertebral body then anterior bone destruction.
Ultrasonography is useful in SA for identifying the presence of deep effusions and in OM for subperiosteal collections, but cannot differentiate between purulent and non-purulent material.7,36 Ultrasonography may also be used to distinguish infection from other causes of similar symptoms or to direct fine-needle aspiration.37
Computerised tomography is most valuable for guided procedures, such as aspiration or drainage of the infected bone or joint.38 It effectively demonstrates air, sequestra and cortical destruction in chronic OM,36 but gives non-specific results in discitis.
Microbiological investigation
Identification of the pathogenic organism by culture should be attempted, with samples preferably taken prior to starting antibiotic therapy, as positive identification of the causative organisms allows targeted antibiotic therapy. Blood cultures, joint fluid (from aspiration), periosteal pus or bone biopsy can all be used. Samples from the infected bone or joint require an invasive procedure but are more likely to be positive (40–50% positive) than blood cultures (9–22% positive).15,27 Yield is generally not high for identification of a bacteria in children with OM,27 as, unless therapeutic operative intervention is required, bone biopsy is infrequently necessary for diagnostic reasons alone.
New molecular techniques including polymerase chain reaction (PCR) and broad-range 16s ribosomal deoxyribonucleic acid PCR39,40 have established the basis for more rapid and sensitive microbiological diagnosis,18 although these methods currently do not provide information on specific organism antibiotic resistance profiles.
Blood cultures (a minimum 4-ml aerobic culture sample in older children, 2 ml in specific neonatal aerobic bottle)41 should therefore be taken and, where available, samples from infected bones or joints placed in a sterile universal container and sent for culture and sensitivity testing. Older reports suggesting an increase in K. kingae recovery is gained from inoculating synovial fluid or bony exudates directly into blood culture bottles have not been replicated in UK practice.17
K. kingae is detectable using new PCR techniques from cultures where conventional direct plating of specimens on solid media has been used.18,19
Surgical management
There is little current high-quality evidence on which to base current surgical practice.
Osteomyelitis
Surgical drainage in acute OM is indicated if the patient is not responding to antibiotics after 48–72 hours (although this may be because of resistance) or if there is radiological evidence of a substantial pus collection.7 Best practice is to immobilise any surgically treated limb or focus of infection. Occasionally, where a soft tissue or subperiosteal collection is clearly demonstrated by ultrasonography or MRI, needle aspiration can be performed prior to starting intravenous (i.v.) antibiotics. When performed, the procedure should be carried out under sterile conditions. If there is bony destruction or pus aspirated, surgical debridement is usually required. With only early radiographic signs, conservative i.v. antibiotic therapy may suffice.
Historically, the role of surgery is poorly defined. Cole et al.42 identified three groups of patients. In the group of patients aged > 1 year who presented within 48 hours, antibiotic therapy alone was sufficient. In the group aged > 1 year, 5 days after the onset of illness, patients usually required surgery and possibly multiple procedures. In infants aged < 1 year, in whom the exact diagnosis was difficult to make, a single operation and antibiotic therapy usually sufficed. In current practice, the relative roles of bacterial virulence and host age and immunity are unclear. More invasive surgery appears more common when bacteria have specific virulence genes, for example PVL.22 Although most children recover rapidly with simple medical management, others require repeated debridement.
Septic arthritis
In SA, prompt drainage and washout of the affected joint (either arthroscopic or open) is advocated by some for both diagnostic and therapeutic purposes as the articular cartilage is damaged early.7 The role of surgery in the treatment of SA is, in fact, poorly defined except in relation to the hip, where prompt surgical drainage is absolutely necessary. Open capsulotomy to allow continuing drainage of septic material is advocated, and, if the arthrotomy does not provide turbid material, drilling the femoral neck may decompress a proximal femoral OM. The anterior approach is preferred as this also allows open reduction of any displacement of the femoral head.
The indications for surgical drainage of septic joints other than the hip remain controversial. Where there is a large effusion, drainage is usually advocated, although in some joints arthroscopic irrigation may be appropriate, such as the knee or ankle. However, with arthroscopic treatment joint visualisation is less complete. Overall, for joints other than the hip, aspiration, irrigation and i.v. antibiotic therapy is the preferred first line of treatment. If the patient fails to respond, then the joint should be surgically drained, usually by formal open arthrotomy rather than arthroscopic drainage.
Medical management and antibiotics
Current evidence for how to initiate treatment
Intravenous antibiotics are started empirically as soon as the clinical diagnosis of acute OM or SA is made, as delaying therapy until the bacterium is identified increases the risk of complications. In SA, where urgent surgery is indicated, a widespread pragmatic approach has been to start antibiotics following surgery unless it will take > 4 hours to get to theatre. As soon as organisms are isolated, antimicrobial treatment should be adjusted and optimised. In subacute OM with no systemic reaction, oral antibiotics can be used from the start.
Although there has not been a definitive randomised controlled trial (RCT), a number of observational and retrospective studies in the literature show that several different antibiotic regimes have been effective in treating acute haematogenous OM in children, including the use of beta-lactam and macrolide antibiotics.9
The initial antibiotics should always include potent cover against MSSA and GAS, and in younger children against K. kingae, although the choice will vary according to the age of the child, route of infection and local resistance patterns.8 Recent data from the USA suggest increases in resistant S. aureus (MRSA), but these data have not been replicated in UK cohorts.43 Activity against H. influenzae type b is essential in children who have not been fully immunised against it.
Switch to oral antibiotics and total duration of treatment
Currently there is no international and little UK consensus regarding the route or duration for antibiotic treatment of acute OAI in children.
Oral switch
Sequential i.v. and oral therapy has become usual as it is less inconvenient and painful for the patient, is associated with fewer complications and is cheaper than longer courses of i.v. therapy alone.3,7,8 There is no current evidence to aid the clinical decision of when to switch from i.v. to oral therapy, although this practice is widely accepted and usually occurs when the patient has shown a marked clinical improvement.9 A Canadian systematic review of short- (≤ 7 days) versus long-course (> 7 days) parenteral antibiotic treatment for acute haematogenous OM in children primarily due to S. aureus found no difference in the overall cure rate after 6 months between short- and long-course parenteral antibiotic therapy.44 A recent retrospective cohort study of 1969 children in the USA found that early switch to oral therapy (median 4 days) was as effective as prolonged i.v. treatment,45 a finding also suggested in a smaller retrospective study of 186 children with SA.46 The laboratory or clinical parameters that would determine the decision to switch to oral therapy remain undefined.
Most clinicians continue i.v. antibiotics until the child shows clinical improvement, if afebrile and oral fluids and medication can be established. Additionally, observing a decrease in inflammatory markers such as WBC, CRP and ESR is thought to be of value.3 Studies have shown that serum CRP level decreases more rapidly than ESR in children recovering from acute OM, and that children with a raised CRP level are more likely to have symptoms or extensive radiographic abnormalities.28,47,48 A recent Finnish clinical trial47,49 reported apparently good long-term results and apparently no failure rates using CRP as the biological marker of infection.
Failure to improve necessitates repeat blood culture, additional imaging for metastatic infection, assessment for deep-vein thrombosis and consideration of unusual pathogens such as PVL S. aureus or Fusobacterium.
No UK consensus currently exists to guide the criteria for oral switch for use in clinical practice or a clinical trial, which will be determined as part of this feasibility study.
Total duration of antibiotic therapy
The suggested duration for parenteral antibiotic treatment ranges from 3 days up to 6 weeks, resulting from several, mainly observational, studies with a relatively poor level of evidence.9,50 In the past, the overall duration of antibiotic treatment has been considered an important factor to improve outcome and reduce relapse. Several paediatric textbooks recommend at least 4–6 weeks of treatment.3,51
Although there are encouraging data from a recent clinical trial in Finland47,49 and from other review papers and case series, no recent formal RCT has been conducted to show good evidence for shorter courses of parenteral antibiotic treatment. There are a number of reasons why the recent Finnish data may not be directly applicable to practice in the UK or other countries.52 Some historical observational studies showed an association between a short duration of antibiotic therapy and 15–19% poor outcome or relapse with courses of ≤ 3 weeks.53–55
Currently there is no consensus about the route or duration for antibiotic treatment of acute OM in children.
Oral antibiotic choice and dose
Many different regimens are used as oral therapy following switch from oral antibiotics, including co-amoxiclav, flucloxacillin and clindamycin. Although flucloxacillin and clindamycin have good oral bioavailability and excellent tissue penetration, both drugs have to be given orally four times per day and both have poor taste and therefore poor drug adherence of the suspension in small children.56 Although clindamycin rarely leads to C. difficile disease in children, there is no current evidence or consensus regarding oral antibiotic choice that will be acceptable to children and parents in terms of both palatability and dose frequency.
Continuation of intravenous antibiotics for more than 2 weeks
Complex disease requiring continuing i.v. therapy poses problems of vascular access, hospitalisation and schooling. Most children will require central or peripherally inserted central venous long line [peripherally inserted central catheter (PICC)] insertion for long-term antibiotic treatment. Delivery of subsequent care is either in hospital or at home, dependent on local services and the ability to provide outpatient parenteral antibiotic therapy (OPAT), although OPAT services for children are not yet well developed in the UK. Central venous lines or PICCs and OPAT has attendant risks, with 3–11% of central venous line-associated infection noted in the USA.57,58
Additional or second-line antibiotics for complex disease or where resistant pathogens are identified
When cases are complex, additional antibiotics may be advised by local microbiologists, clinical infectious diseases specialists or national guidelines, for example PVL-positive S. aureus infection.59
Complications
Deep-venous thrombosis and thromboembolism have been seen in up to 30% of children with OM and are associated with a higher risk of disseminated infection.60 In addition, joint stiffness, limb shortening, dislocation (acutely neonates) and avascular necrosis of affected epiphysis may occur.
Routine follow-up allows most children with simple disease to be discharged without the need for long-term care or further assessment of growth or function.
In the context of clinical audit or clinical trials, outcome measures may include length of stay in hospital, total length of therapy, operative procedures required, as well as formal assessment of growth and function.
Currently accepted clinical equipoise
At this time, all clinical co-investigators have agreed in writing that they agree there is no international and little UK consensus regarding the route or duration for antibiotic treatment of acute bone and joint infections in children, and that there is also no clear evidence to aid the clinical decision of when to switch from i.v. to oral therapy.
