There are no uniform clinical criteria for the diagnosis of prosthetic joint infections. Most patients experience a long, indolent course of infection characterized by steadily increasing joint pain and the occasional formation of cutaneous draining sinuses. A minority of patients have an acute fulminant illness associated with high fever, severe joint pain, local swelling, sinus tract, and erythema. Patients with late-onset infections due to hematogenous seeding can present with this acute onset of symptoms in one or several previously well-functioning joints. The pattern of clinical presentation is determined largely by the nature of the infecting microorganism (i.e., the symptoms are more prominent in S. aureus infections compared with S. epidermidis). Infection must be differentiated from aseptic mechanical problems. Constant joint pain suggests infection, whereas mechanical loosening commonly causes pain only with motion and weight-bearing. Nevertheless, it is often difficult to differentiate delayed-onset infection from aseptic joint loosening of hip or knee prostheses.

Persistent elevation of the erythrocyte sedimentation rate (ESR) suggests infection, but is neither very sensitive nor very specific since it may be due to many other causes. The same is true for leukocytosis or C-reactive protein (CRP). The combined measurement of CRP and the ESR seems to have an acceptable negative predictive value, although it does not completely rule out a prosthetic joint infection. These two parameters are also more appropriate to measure as follow-up during therapy than for diagnosis. The role of procalcitonin in nonbacteremic prosthetic infections has to be further investigated. A study performed at our center did not prove a diagnostic or follow-up value of procalcitonin in localized prosthetic joint infections (19). In the future, the use of other inflammatory markers in the diagnosis of infection will certainly emerge of which the serum leukocyte esterase warrants further investigation, for example (20).

A plain radiograph can display abnormal lucencies (>2 mm in width) at the bone-cement interface, changes in the position of prosthetic components, cement fractures, periosteal reaction, or the motion of components. Periosteal new bone formation is suggestive of infection, but infrequently present as is the presence of fistula between the joint and soft tissue. When both the distal and proximal components of a prosthetic joint demonstrate radiographic abnormalities, infection is more likely than simple mechanical loosening. The sensitivity and the specificity of radiographic anomalies have been reported as 73% and 76%, respectively (21). Magnetic resonance or computed tomography techniques for evaluating prosthetic joints for infection are of little help because the metal present in prostheses interferes with these images. Radioisotopic scans demonstrate increased uptake in areas of bone with enhanced blood supply or increased metabolic activity, but this does not help to diagnose true infection because increased uptake is routinely observed around normal prostheses for several months after arthroplasty. Smith et al. investigated the usefulness of bone scintigraphy after a mean period of 3 years between knee replacement surgery and the onset of knee pain. However, the pattern of isotope uptake in the abnormal studies was not specific enough to differentiate aseptic from septic loosening (22). This result was confirmed in a report of 144 patients who underwent revision hip arthroplasty; bone-gallium imaging offered no additional diagnostic advantage over hip aspiration (23). According to Bernard, the sensitivity and specificity of bone scintigraphy is at best 76% (21). Only limited data are available for positron emission tomography (PET). In a report of 35 patients with painful hip replacements, PET scan performed similarly to scintigraphy and was less sensitive and more specific than
conventional bone radiography (24). In conclusion, a normal bone scan is generally useful to exclude the need for surgical intervention aimed at correcting loosening or infection of the joint (25).

Several identical positive microbiologic cultures identifying the same microorganism(s) are currently the gold standard to confirm the clinical suspicion of infection. As a result, the diagnosis of prosthetic joint infection always requires obtaining samples of joint fluid or tissue (26). The most important step is to isolate the offending organisms so that the appropriate antimicrobial therapy can be chosen. The importance of the identification of the pathogen cannot be overemphasized. Alternatively, isolation can be achieved by blood cultures, sonication of explanted prostheses (27), or by direct biopsy from the involved bone, since joint fluid aspirates may be falsely negative (28). Whenever biopsies or aspirations are done, the samples should be processed for aerobic and anaerobic cultures. Samples for mycobacterial and fungal cultures should be taken and processed if commonly cultured microorganisms are not present and if the clinical features are compatible. Often, culture growth has to be extended beyond the standard incubation period of 5 days. Cultures may be negative because of prior antimicrobial exposure, a low number of organisms, an inappropriate culture medium, fastidious organisms, or prolonged transport time to the microbiology laboratory. Some microorganisms, such as Abiotrophia spp. or staphylococcal small-colony variants (SCV), are inherently difficult to detect (29). In contrast to the normal staphylococcal phenotype, SCVs are characterized by an enhanced ability to persist intracellularly, fastidious growth characteristics, nonpigmented and nonhemolytic pinpoint colonies after 24 to 72 hours of incubation on blood agar, and the absence or reduction of biochemical reactions (e.g., mannitol salt agar negative) (29). SCVs are rapidly overgrown and are easily missed when the normal S. aureus is present because they divide about nine times slower than normal-phenotype S. aureus. Isolates suspicious for S. aureus SCVs, which may give a false-negative coagulase test, should be confirmed as S. aureus by testing the species-specific genes or by diagnostic sequencing of 16S rDNA sequences (29).

Eubacterial polymerase chain reaction (PCR) has usually only little sensitivity and is still relatively expensive, which excludes its routine application. In polymicrobial colonization or infection, its interpretation can be difficult. Moreover, it does not provide information about antibiotic resistance, except for genes coding for methicillin resistance (30). However, specific or multiplex PCR is beneficial in special circumstances when very slow- or difficult-growing bacteria are suspected, such as Kingella kingae, Brucella spp., Coxiella burnetii, Bartonella henselae, Mycobacterium tuberculosis, or M. ulcerans.

Tissue specimens should also be submitted for histopathologic study, as special staining techniques may reflect unusual or slow-growing microorganisms. A histopathologic examination showing acute inflammation has a >80% sensitivity and >90% specificity for the diagnosis of infection (31). However, the results can be dependent on appropriate sampling of the tissue harboring the infection and the expertise of the pathologist. Not all centers will have pathologists who are experienced in this type of histopathologic analysis.