Is Early Diagnosis of Necrotizing Fasciitis Important?

This edition of the Journal contains a paper by Su et al. reporting a cohort of patients in Taiwan who were ultimately discharged with a diagnosis of necrotizing fasciitis (NF).1 The authors specifically wanted to test the ability of the laboratory risk indicator for necrotizing fasciitis (LRINEC) score to detect which patients would probably die or require amputation of a limb on admission. Wong et al. had developed the LRINEC as a tool to assist early diagnosis of NF.2 Su et al. divided their cohort by LRINEC score as recommended by Wong and concluded, both from a direct between-group comparison of the risk of mortality and amputation and using receiver–operator characteristic (ROC) curve analysis, that an LRINEC score of 6 or greater was indeed associated with an increased risk of both amputation and death. This is a potentially useful addition to our understanding of which patients are most at risk of poor outcome with NF, but it remains to be seen how useful this predictive ability will be in clinical practice. Necrotizing fasciitis is part of a range of fulminant necrotizing infections that are characterized by destruction of skin and soft tissues down to and including the subcutaneous fascia, but sparing the underlying muscle. They are usually the result of infection with a synergistic mixture of bacteria. First described in the scrotum and penis by Fournier in 1883, and subsequently by Meleney in a series of patients with streptococcal disease, prompt diagnosis and immediate aggressive management are accepted as the keys to optimal outcome. These infections are rare, but are commonly associated with systemic toxicity and death. Recent clinical series suggest the current expected mortality is approximately 15–30% and probably depends on many factors including severity, time and the availability of appropriate local resources. Wong et al. proposed that early diagnosis might be made according to the LRINEC, which uses six easily obtainable and rapidly available blood tests (white cell count and concentrations of C-reactive protein, haemoglobin, sodium, glucose and creatinine). These authors retrospectively validated the diagnostic accuracy of the score using ROC curve analysis and suggested a cut-off value of 6 or greater. The positive predictive value (PPV) of an accurate diagnosis was 92% and the negative predictive value (NPV) 96%. Wong suggested that early diagnosis might improve outcome by facilitating earlier definitive intervention in the form of radical surgical debridement of all dead tissue, appropriate antibiotics and intensive general support measures. The results of the study published in this issue of the Journal support this suggestion by showing some association between higher LRINEC scores at admission and those with poorer clinical outcomes. The association is, however, not as impressive as the diagnostic accuracy as estimated by Wong. Chen's results suggest that an LRINEC score of 6 or greater will predict approximately 64% of the cases that will ultimately die (sensitivity), whereas the PPV is only 21% – that is, only 21% of all those with a score of 6 or greater will ultimately die. This is similar to a mortality of 11% in those who score less than 6. Similarly, the same scores predict 66% of those who will require an amputation; the PPV is 36% and the amputation rate in those scoring less than 6 is 17%. These figures suggest the LRINEC is only moderately predictive of outcome in NF, but this may be missing the potential effect of LRINEC on clinical outcomes. In this retrospective cohort, it is reasonable to assume that the physicians making the early assessments and diagnoses in these patients were not aware of the LRINEC and therefore did not use these scores to achieve early diagnosis and prompt referral for operative management, aggressive antibiotic therapy and pre-emptive intensive care support. In fact, the cohort reported by Su has done rather well in comparison with the published experience and it worth noting that 67% of these patients were diagnosed correctly with NF within 3 h of presentation to the emergency department. It may be that earlier diagnosis than this is either a practical impossibility or is of no clinical importance. We cannot be certain, therefore, of any influence of the LRINEC score on improving outcome and a carefully targeted prospective research effort is required. Several other tools have been suggested to improve the speed and accuracy of diagnosis. Most experts in the field agree that the first requirement is a high degree of suspicion. Although comprehensive risk factors for NF have not been identified, there are strong associations with immunocompromise, diabetes mellitus, i.v. drug use, obesity and poor self-caring. Some clinical features at presentation are highly specific for NF (or a least a necrotizing infection of some type), but are not particularly sensitive, being present in only 10–40% of the patients. The diagnosis may therefore be missed. Such features include tense oedema in the surrounding tissues, satellite areas of necrotic skin, skin blisters, subcutaneous gas and pain that is out of proportion to the apparent lesion.3 The absence of both a high white cell count and low serum sodium together (>5400 cells/mm3 and <135 mmol/L) is useful to rule out NF (NPV 99%), but the presence of both is a poor predictor of NF (PPV 26%).4 Similarly, the presence of subcutaneous gas on plain radiographs is highly suggestive of a necrotizing infection, but the absence of gas does not rule out NF. The place of other imaging techniques, including computed tomography and magnetic resonance imaging remains unclear and such tools either may not be readily available in the facility of first presentation, or actually delay definitive treatment. Microscopic tissue appearance on frozen section may be indicative of NF, but has not been widely accepted and again relies heavily on a high clinical suspicion of NF. In summary, the association shown by Su et al. between LRINEC score at presentation and poorer clinical outcomes reflects the unsurprising fact that those with more severely deranged biochemistry are sicker and will probably do poorly. This, in turn, may only be a matter of how long the infective process has been under way before presentation. The search for a diagnostic tool that rapidly and reliably identifies patients with NF and also allows time for effective therapy to modify clinical outcome, remains an active area of investigation. Any useful diagnostic tool is either likely to continue to require a high degree of clinical suspicion or it will not be applied. In practice, many cases of NF are diagnosed on the operating table at the first attempt to debride the necrotic tissue. In Australia and New Zealand, each individual hospital surgical department is not likely to see even one case of NF each year. In the small number of tertiary referral centres that see multiple cases annually (usually those with hyperbaric oxygen capability), referral is often made after at least one operative debridement. Awareness of diagnostic scores such as the LRINEC may prompt preoperative discussion with tertiary centres where experienced surgeons, intensivists, infectious disease and hyperbaric specialists may be useful in guiding immediate operative and ancillary management, as well as facilitating prompt postoperative transfer if required.