Necrotizing fasciitis (NF) is a rapidly progressive soft tissue infection requiring prompt diagnosis and surgical treatment. The differentiation of NF from nonnecrotizing (non-NF) fasciitis, such as cellulitis or abscess, may be difficult. The “hard” clinical signs usually associated with NF, including hemodynamic instability, crepitance, skin necrosis, bullae and gas on radiograph, often are not present in patients with NF. Magnetic resonance imaging and frozen-section tissue biopsy can be helpful but may not be readily available for rapid diagnosis.

Wall and associates studied the use of an elevated white blood cell (WBC) count (greater than 15,400 per mm³ [15.4 × 10⁹ per L]) or a serum sodium level less than 135 mEq per L (135 mmol per L) as a predictive model to distinguish between NF and non-NF patients. A retrospective record review of 31 patients discharged from the hospital with a diagnosis of NF and 328 discharged with a diagnosis of non-NF included the admission WBC count and the serum sodium concentration (corrected if hyperglycemia was present). The NF patients had a significantly lower mean admission systolic blood pressure and a higher mean admission respiratory rate. On physical examination, NF patients significantly more often exhibited tense edema, bullae, purplish skin discoloration, and sensory or motor deficit. Skin necrosis or gas on plain radiography was not found significantly more often among the NF patients. No patient was reported to have crepitance.

This model was positive in 90 percent of NF patients and 24 percent of non-NF patients. This finding represents a sensitivity of 90 percent and a specificity of 76 percent. The positive predictive value was 26 percent, and the negative predictive value was 99 percent.

The authors conclude that a predictive model based on a WBC count greater than 15,400 per mm³ and a serum sodium level less than 135 mEq per L is useful in distinguishing NF from non-NF. Less than one half of the NF patients demonstrated hard clinical signs, and none had crepitance. A model that assists the clinician when these signs are not present will encourage accurate diagnosis of NF, resulting in faster initiation of treatment. More significantly, because of the high negative predictive value of the model, patients without hard clinical signs and a negative model result are unlikely to have NF.