Diagnostic performance of MRI and CT in diagnosing necrotizing soft tissue infection: a systematic review

Objective To systematically review the accuracy of MRI and CT in diagnosing necrotizing soft tissue infection (NSTI). Methods MEDLINE and Embase were searched for original studies which reported the diagnostic performance of MRI or CT in detecting NSTI. Individual study quality was assessed using the QUADAS-2 tool. Sensitivity and specificity of MRI and CT were calculated and, if supported by data from at least two studies, pooled using a bivariate random-effects model. Results Six MRI studies and 7 CT studies were included. There were no major concerns with regard to study quality and applicability. The included studies used multiple diagnostic criteria, with sensitivities and specificities of both MRI and CT ranging between 0 and 100%. T2 hyperintensity of deep fascia was the most commonly used diagnostic MRI criterion (5 studies), yielding a pooled sensitivity of 86.4% (95% confidence interval [CI] 76.1-92.7%) and a pooled specificity of 65.2% (95% CI 35.4-86.6%). Presence of gas was the most commonly used diagnostic CT criterion (3 studies), yielding a pooled sensitivity of 48.6% (95% CI 37.1-60.2%) and a pooled specificity of 93.2% (95% CI 73.3-98.5%). Conclusion T2 hyperintensity of deep fascia at MRI has high sensitivity and moderate specificity in diagnosing NSTI. Presence of gas at CT has low sensitivity but high specificity. A combination of diagnostic criteria may improve diagnostic performance, but this needs further investigation.