Introduction

Septic arthritis is defined as a bacterial infection of the synovium and joint space. It is a relatively uncommon condition that may occur at any age being more common in children. In the pre-antibiotic era the mortality of all joint infections was 20% and morbidity in the region of 50%. Joint infection leads to rapid and irreversible damage to the articular cartilage. Any delay in the treatment of a septic joint will result in a poor outcome. While it is an infrequent presentation in the elbow, early recognition, diagnosis and treatment are essential to achieve a satisfactory outcome. This requires the clinician to have a high index of suspicion of sepsis in patients with predisposing conditions. In addition, infection can obviously follow any surgery to the elbow and is a particularly troublesome complication of elbow arthroplasty. This has been dealt with in Chapter 44 and will not be covered further in this chapter.

Epidemiology

In developed countries septic arthritis has shown no change in its incidence¹ despite advances in antimicrobial therapy. Among children in Sweden the annual incidence is 10 per 100 000 children.² In a separate study from the Mayo clinic over a 15-year period, the elbow accounted for 6% of all septic joints in the adult. To put this in context, however, the number of patients presenting to that unit with septic arthritis unrelated to previous surgery was only 10 new adult cases per year, thus emphasizing the uncommon nature of elbow sepsis. Most of these cases were over the age of 50 years with an equal sex ratio.³

The incidence in developing countries, particularly in Africa, is considerably higher.^4,5 In contrast to the experience in Europe and North America, Goldschmidt and Hoffman⁵ reported a 100% increase in presentation of septic arthritis in Cape Town between 1983 and 1988, with a male to female ratio of 2:1. They also noted a seasonal variation with a 30% increase in incidence in the cooler months. This seasonal variation is the converse of their finding with other musculoskeletal infections which were 50% higher in the summer months. Other reports also suggest that musculoskeletal infections are commoner in warmer and more humid conditions.⁶

The presentation of septic arthritis may be monoarticular or polyarticular. The polyarticular form accounts for 10% of all presentations and, in adults, half of these are in rheumatoid arthritis patients.¹

Aetiology and pathogenesis

There are four routes by which bacteria can spread to the elbow joint:

Pathology and microbiology

Bacterial inoculation within the joint will cause infection by bacterial proliferation. This causes an inflammatory reaction within the synovium with an increased blood flow. Increased capillary permeability allows fluid and protein leakage with the attraction of polymorphonuclear leucocytes and phagocytes. This exudate of bacteria, both living and degraded, white blood cells and proteins enters the synovial cavity resulting in a septic effusion. In cases of haematogenous spread, more than one joint may be involved.

This cascade with an infective effusion is distinct to that seen in a reactive arthritis. In reactive arthritis there are no live bacteria within the joint cavity. An effusion occurs either in response to immune complexes or bacterial cell wall products. Immune complexes giving rise to reactive arthritis may be seen in non-bacterial infections. This is seen in several viral infections such as rubella and mumps. Some bacterial infections that spread haematologically may cause both infective and reactive effusions in the same patient. Neisseria gonorrhoeae can present with a polyarticular pattern of effusions of which only 20% of the affected joints may be infected and the remainder reactive. Rarely protozoa and fungi have been implicated in joint infection, most of these are, however, polyarticular and reactive in nature.

There are thus three patterns of bacterial infection that may give rise to a joint effusion: monoarticular and polyarticular infective effusions and reactive effusions. The importance of distinguishing between reactive and infective joints is obvious, but can be difficult, especially in systemic bacterial infections where these types of effusions can coexist. Joint infections can be caused by a wide variety of different bacteria. Tuberculous infections, however, present differently and have management regimens that set them apart from other bacterial infections. These infections will thus be considered under a separate heading.

The commonest infective agent in joint sepsis in both the adult and child is Staphylococcus aureus (55%) with streptococcal infection accounting for 18% of infections, being commoner in patients over the age of 60.^12,13 Meticillin-resistant S. aureus is increasing in frequency, causing a third of staphylococcal infections in one study. The affected patients were older and had more comorbidities and a higher 6-month mortality rate.¹⁴ N. gonorrhoeae infection is reducing in incidence but remains the commonest cause of septic arthritis in the young adult.^15,16 A wide number of other infective agents have been described. There is, however, an association of different causative bacteria with age (Table 48.1).

Table 48.1 Causative bacteria in different age groups

Natural history

The introduction of sulphonamides in the 1930s and the proliferation of antibiotic agents after 1940 led to a dramatic improvement in the outcomes of patients with septic arthritis. Before the antibiotic era there was 20% mortality and 50% morbidity from joint sepsis.¹⁷ However, even in the 1980s a 13% mortality rate was reported following disseminated joint infection in children.¹⁸ This therefore remains a serious condition particularly in the young, old and the immunocompromised.

The presence of active infection in the joint rapidly causes irreversible changes. In animal models, maximal arthritic symptoms occur 2 days following the inoculation of bacteria into a joint and at 7 days irreversible changes have taken place.¹⁹ The damage to the articular cartilage has historically been attributed solely to the effect of increased pressure within the joint resulting in the surgical drive to drain the joint as soon as possible. This has remained unquestioned until relatively recently. Other causes of high joint pressure are seen in effusions and haemarthrosis and yet the rapid destruction of the joint is not seen in these conditions. It is self-evident, therefore, that other factors which are not mechanical must have a part to play in the joint damage.

The effect of the inflammatory cascade releases proteolytic enzymes and cytokines into the joint. These have a rapid and devastating effect on the articular cartilage. Articular cartilage when exposed to fibronectin fragments undergoes proteoglycan depletion within 7 days.²⁰ TNF-α, interleukin-1 and interleukin-6, which are released into the joint in sepsis, are inflammatory and cause rapid articular surface damage.^21,22 These changes have been observed in the absence of mechanical factors. These studies demonstrating the rapid degradation of articular cartilage within a few days in animal models correlate with the clinical picture seen in humans and emphasize the need to diagnose and treat this condition early.

Clinical presentation

The patient will usually present with a painful and swollen elbow and a general feeling of malaise. In all patients presenting with these signs there should be a high index of suspicion for sepsis. Painful movements leading to a loss of function is also an early sign. As the effusion enlarges, the elbow is held in 80° of flexion, which is the position of comfort allowing the maximal joint volume.²³

Classical clinical signs will include a red, warm joint with an effusion together with pain and limitation of movements particularly of flexion and extension. The patient is also usually systemically unwell and pyrexial. It should be remembered, however, that in the majority of cases patients will be either young and immunocompromised or be experiencing an inflammatory arthropathy. These patients may not display the classic clinical signs.

In the young, any pyrexia is usually marked and the child is plainly unwell. Under the age of 1 year local signs may be minimal, again making diagnosis difficult.²⁴ Patients who have rheumatoid arthritis or who are immunosuppressed may present with rather vague symptoms of feeling unwell with few local clinical signs other than an effusion. These patients often have less redness and pain and more movement than one might anticipate with a septic joint. Again a high index of suspicion is required to make the diagnosis in these patients.

The differential diagnosis is that of a hot swollen joint. This will include other sepsis such as an olecranon bursitis or cellulitis, inflammatory arthritis, crystal arthritis, trauma and haemarthrosis. In children the presence of osteomyelitis can mask the diagnosis of a septic joint and lead to a delay in diagnosis.¹¹ This may account for the significant number of poor results in children treated for osteomyelitis.¹⁰

Investigations

Patients presenting with a red swollen elbow should have routine blood investigations undertaken and a screening X-ray. Blood tests will usually reveal an elevated erythrocyte sedimentation rate (ESR) to at least 50 or more and an elevated C-reactive protein (CRP), which may be more sensitive in patients with inflammatory arthropathies. The white cell count may be raised to greater than 12 000/mm³ with a 40–50% polymorphonuclear leucocytosis. Blood cultures are also recommended since they may be positive in up to 50% of cases.

Plain radiographs are usually normal or show signs of an effusion. While radiographic signs of sepsis usually appear after 2–3 weeks, X-rays are useful by way of the differential diagnosis and to give a reference point for future radiological examination. Other imaging is possible but rarely contributes significantly in the acute situation. Ultrasound and magnetic resonance imaging are accurate in diagnosing an effusion, however, neither can distinguish between septic effusions and other causes.²⁵ The technetium-99m phosphate three-stage bone scan is accurate in identifying increased uptake in septic joints in nearly all cases which may contribute to diagnosis but at the expense of delaying the onset of treatment. Likewise indium-111 labelled leucocytes have a sensitivity of 80% in the diagnosis of sepsis²⁶ but this investigation takes 48 hours, which again leads to an unacceptable delay.

Joint aspiration is the most important confirmatory investigation and should be performed as soon as the diagnosis is in question.¹⁵ Aspiration should be performed using aseptic technique, both to reduce the risk of patient infection in the non-septic joint and reduce the risk of bacterial contamination of the specimen. Septic effusion is often viscous and a wide-bore (14 G) needle is recommended. Local anaesthetic infiltration of the skin is thus advisable, though the anaesthetic should not be injected intra-articularly prior to aspiration since local anaesthetics are bacteriostatic.

The recommended portal for aspiration is a lateral approach with the needle insertion being at the central point of the equilateral triangle formed by the lateral epicondyle, the olecranon process and the radial head. The radial head is often not palpable but its position may be surmised by drawing the base of the triangle between the other two easily palpable landmarks (Fig. 48.1).

Figure 48.1 Palpate the lateral epicondyle (LE) and the tip of the olecranon (O) and draw a line between these landmarks. Complete an equilateral triangle. The radial head (RH) lies at its apex. The central point (X) marks the joint line and the point of aspiration.

Inspection of the aspirate reveals a turbid or opaque appearance of moderate viscosity. Specimens should be sent for Gram staining, culture and sensitivity, white cell count and crystals. Protein and glucose analysis may also be useful. The white cell count is usually over 100 000/mm³ with 75% of polymorphonuclear leucocytes seen. Crystal analysis is negative. Protein analysis is near normal and glucose analysis is less than 50% of the blood value. These latter two tests may distinguish from inflammatory causes where the protein is often markedly raised and the glucose is only moderately reduced to 75% of the blood level. A positive culture though, as with imaging techniques, may take 48 hours to become available. Thus treatment should be started empirically as soon as the clinical diagnosis of sepsis is made. Antibiotic sensitivity should be sought for all positive cultures.

Treatment

The mainstay of treatment is drainage of the elbow and antibiotic therapy. Drainage may be achieved by repeated aspiration, arthroscopic washout or arthrotomy and washout. The choice between arthroscopic washout and washout via an arthrotomy is largely one of surgical preference, expertise and equipment availability. Septic arthritis caused by a penetrating injury is better treated by arthrotomy and debridement of the tract.

Arthroscopic washout can be performed through either a medial or lateral portal, or both, providing good irrigation of the joint. Consideration needs to be given to a posterior collection particularly in later presentations since this may not drain readily through the above two portals and a third posterior portal may be needed. An arthrotomy is usually performed through a limited lateral approach, although it is important to access all areas of the elbow.

One area of contention currently among orthopaedic surgeons is whether or not drainage of the joint by needle aspiration, with or without needle washout, is a safe treatment technique or whether formal surgical drainage is required. As long as drainage to dryness can be achieved by needle aspiration it would appear that this method is acceptable.²⁷ If the needle drainage is either not satisfactory or incomplete, or if deterioration occurs over 24 hours, formal washout by arthroscopic or arthrotomy methods should be undertaken.^1,27

Antibiotic therapy should be started as soon as the possibility of a septic arthritis has become part of the differential diagnosis. The antibiotic choice before culture and sensitivities are available is based empirically on the most likely infective organism. This judgement can be aided by considering the patient’s age (Table 48.1), known foci of infection (such as urinary tract infection in the elderly) and the Gram stain of the aspirate. High-dose flucloxacillin or a cephalosporin is frequently used. The advice of a microbiologist is recommended (Table 48.2).

Table 48.2 Summary of recommendations for initial empirical antibiotic choice in suspected arthritis¹⁷

Initial treatment should be given intravenously for 7–14 days following which oral antibiotics should be continued for a further 4–5 weeks.²⁸ Monitoring the haematological indices such as the white cell count, ESR and CRP is recommended. The antibiotic regimen should be altered if necessary as soon as the sensitivities are available.

Direct intra-articular antibiotic delivery to a septic joint has also been suggested. There is historical evidence that sterilization of the joint is more rapid in cases where this is undertaken.²⁹ However, there must also be some concern that such treatment can cause a chemical synovitis and thereby result in additional articular cartilage damage. In addition, intravenous administration of antibiotic gives a good delivery of antibiotic to the synovial cavity and this route rapidly achieves therapeutic antibiotic levels. Intra-articular antibiotic delivery is therefore not recommended.

Postoperatively the options are whether to splint or mobilize the elbow. A period of immobilization of 24–48 hours for pain relief has been suggested. However, early active mobilization has the advantage that further exudate can be expressed out of the joint by movement in cases that have had an arthrotomy and, to a lesser extent, an arthroscopic washout. This exudate may contain bacteria, harmful enzymes and cytokines. In addition, Salter et al found additional benefits of early mobilization in preventing adhesions and pannus, improving the nutrition of the articular cartilage and stimulating chondrocytes.³⁰ Early active mobilization is thus recommended.

Since the main harmful effect of sepsis on the articular cartilage is chemical in origin and the result of this damage is catastrophic to the joint, there would be a benefit in directly neutralizing these harmful mediators. Much work has been done in this field in animal models. Neutralizing TNF-α and interleukin-1 with anti-TNF-α and anti-interleukin-1 has the potential to significantly reduce both inflammation and articular surface damage.²¹ Interleukin-4 and interleukin-10 block polymorphonuclear neutrophils in rat models of arthritis using mycobacterial antigen. The reduction in phagocytic activity in response to proinflammatory cytokines may make these agents useful.³¹ Interleukin-12 has been used in modifying the response in septic arthritis in mice infected with group B streptococci. A dose-dependent beneficial effect was found when the interleukin-12 was administered before the clinical signs become obvious.³²

The potential for treatment by the reduction of the harmful effects of the inflammatory process in the septic joint is self-evident. The neuropathic elbow joint may become infected by haematogenous spread to the damaged joint, haemarthrosis, or by direct inoculation if skin breakdown occurs. These patients have an additional challenge of significant instability, which can become worse with the onset of sepsis (Fig. 48.2). Temporary stabilization by the application of an external fixator may aid treatment in these cases.⁷ Wound healing in the grossly unstable elbow is a problem and thus sepsis from haematological spread should be drained either by needle drainage or arthroscopy rather than arthrotomy.

Figure 48.2 Neuropathic elbow (A) demonstrating significant increased instability following infection (B).

Complications

Septic arthritis is a serious condition that still carries a high mortality. The high mortality rate with disseminated infection in children has already been alluded to.¹⁸ Gupta et al reported an 11% mortality rate in a prospective study of 75 patients with adult-onset septic arthritis. Many of these had concomitant disease, 46 patients had rheumatoid arthritis and 11 were intravenous drug users. Interestingly in this series those who had leg ulcers had 38% mortality.³³ Mortality in these cases is due to septicaemia. In all patients and particularly those with an inflammatory arthropathy or those who are immunocompromised, the general effects of sepsis must not be overlooked. Patients can become rapidly severely ill with septic shock. Careful monitoring of the pulse, blood pressure, respiratory rate, temperature, urine output and oxygen saturation is required. An awareness of the local and national guidelines for the management of sepsis is essential.³⁴

Recurrent infection is described though it is not common. Treatment is as for primary infection, though in cases where aspiration with or without needle washout was used as the initial treatment, second time around a more aggressive washout, either arthroscopically or by arthrotomy, should be undertaken.

Late presentation or diagnosis leading to a delay in treatment for 7 days or more leads to irreversible joint changes in the form of chondrolysis. Radiographically there may be periarticular osteoporosis and loss of joint space. In more severe cases there may be joint disruption with subchondral cysts and even destruction of the bony architecture.

Patients in whom there has been failure of the initial treatment or recurrent infection will complain of persistent pain and swelling of the joint with a reduction in the range of movement. Signs of persistent infection including local signs of inflammation and lymphadenopathy may be masked by the antibiotic treatment. Later presentation with pain, a loss of function with reduced movement or even ankylosis is suggestive of chondrolysis. Patients with significant bone destruction may, in addition, have signs and symptoms of instability.

These patients present a surgical challenge. Initially confirmation that the infection has been completely eradicated is essential. Further aspirates can help with this; however, synovial biopsy may be required to confirm that the infection is completely eradicated.

Formal joint debridement can be performed including synovectomy, drainage of any residual fluid collections, intraosseous cyst, removal of any visible pannus, together with any loose or devitalized material. Both free fluid and synovial specimens should be sent for bacteriological and histological examination. In the absence of infection an arthrolysis may be performed to improve the range of motion. Early active mobilization is required thereafter.

If soft tissue surgery fails and the patient remains symptomatic, particularly with pain or less commonly with instability, then salvage surgery may be required. This will take the form of either an arthrodesis or arthroplasty. The arthroplasty may be a resection or total joint replacement. The choice of which procedure will depend on the patient’s age and their expectations.

The timing of any surgery is controversial. A sufficient period should be left to allow the maximal potential recovery in pain relief, swelling and range of motion to take place. In addition, the interval should be sufficiently long to minimize the risk of any subsequent infection from any further surgical procedure, e.g. a total joint arthroplasty. Unfortunately there are no evidence-based guidelines to help the surgeon with this decision. A reasonable time interval would, however, be at least 6 months to allow the joint to settle to a stable state and thereafter to reassess the symptomatology.

The risk of infection following a total elbow arthroplasty in a joint that has previously been septic is certainly higher than in one that has not. Clinical and haematological indices should normalize, but in patients with an inflammatory arthritis this may not occur. The author’s preference is to allow a minimum of 1-year infection-free interval with normal inflammatory markers, before offering a total elbow arthroplasty.

Tuberculous septic arthritis

Tuberculosis is an uncommon cause of sepsis in the elbow. However, the incidence worldwide is significant³⁵ and the incidence in the West is increasing due in part to HIV infection.^36,37 Martini et al^38,39 in a large series of bone and joint tuberculosis identified 652 cases of which 297 were of the peripheral joints. Seventy-four cases affected the upper limb of which 42 involved the elbow. These patients present most commonly with a monoarthropathy. The onset is insidious with symptoms of progressive pain, swelling and stiffness.⁴⁰ While tuberculous infection is often associated with the elderly and infirm, it may present at any age, including in children and young adults.^41,42 Late presentation is, however, common and, in these cases, joint destruction occurs.⁴² A polyarticular presentation with swollen joints, which are sterile in association with tuberculous infection, is recognized (Poncet’s disease). This may be a reactive arthropathy though failure to culture acid-fast bacilli from an affected joint is recognized, leading to some controversy as to the nature of this condition.

The presence of pulmonary tuberculosis should alert one to the diagnosis.⁴³ Aspiration resulting in the culture of acid-fast bacilli should be possible in 75% of cases.²⁹ In cases where there is either a high index of suspicion or where the initial aspirate has been negative or not cultured for acid-fast bacilli, a synovial biopsy should be performed to make the definitive diagnosis.⁴⁴ The characteristic findings on histological examination of the synovium include granulomatous inflammation, Langhans’ giant cells and epithelioid cells, lymphocytic infiltration and caseation.⁴⁵ Later, radiographic examination may show a periosteal reaction and joint destruction.^38,42

Patients who present with a possible tuberculous joint should undergo either an arthroscopic or open drainage and washout with a synovial biopsy. The washout may improve the range of movement.⁴⁶ In late presentations and more advanced cases, therapeutic synovectomy and debridement may be required. Treatment with anti-tuberculous chemotherapy is the mainstay of management and is highly effective.⁴²

Martini et al³⁹ described the outcome of non-operatively managed tuberculous arthritis of the elbow. In 42 cases, 19 had a range of motion of more than 70°, 27 had what they described as a useful range of motion and the remainder had ankylosis. They also found rotation to be restricted.⁹ For patients with residual pain formal arthrodesis may be performed.⁴⁷

Functional results are related to the radiographic stage at presentation and not the duration of symptoms or the range of motion at presentation.⁴² Treatment for patients who have residual symptoms following successful treatment of the infection follows the same rationale as the treatment plan for the poor outcomes of acute infection. Initially debridement, synovectomy, radial head excision and arthrolysis may be considered. Arthrodesis is an option for patients who have residual pain and are young or high demand.⁴⁷ Total elbow arthroplasty may be performed for either pain or in patients who require movement (Fig. 48.3).

Figure 48.3 End stage of a tuberculous septic elbow joint leading to a painful near-ankylosed joint. (A) Postoperative radiograph following arthroplasty. (B) Note the ulnar wire for an infraction in the ulna.

Conclusions/personal view

Septic arthritis of the elbow is a relatively uncommon condition. Delay in diagnosis and treatment leads to irreversible changes in the elbow with chondrolysis, following which treatment options are limited. The most likely patient presenting with this condition will be either one who is immunocompromised or one who has an inflammatory arthropathy. The key to early diagnosis in these cases is to consider the possibility of infection since the classic clinical signs of a red, warm and painful joint that one would usually associate with infection and may be absent. Complex investigations are unnecessary in nearly all cases and can lead to a delay in treatment.

Aspiration has been demonstrated to be a safe treatment option. However, it is important that the joint is aspirated to dryness and that any new collection is drained. Thus if there is uncertainty whether the aspiration is adequate, formal early washout either by arthroscopy or limited arthrotomy should be performed.

Salvage surgery for cases that have developed chondrolysis who present with pain and stiffness have been treated successfully by the author with arthroplasty. The timing of this surgery is an inexact science. Sufficient time should be allowed, however, for symptoms to settle. Added to this, an infection-free interval prior to surgery is essential. This author has used a 1-year infection-free period without infection complicating the arthroplasty.