The diagnostic accuracy of Gram stain on formalin-fixed paraffin-embedded sections of skin tissue in the diagnosis of bacterial skin infection.

BACKGROUND AND OBJECTIVE: To evaluate the sensitivity, specificity, and likelihood ratios of Gram stain on formalin-fixed, paraffin-embedded (GS-FFPE) sections of skin in diagnosing bacterial skin infection. METHODS: We reviewed a retrospective series of skin specimens reported at our institution wherein histopathological assessment included Gram stain and fresh tissue was concurrently submitted for microscopy and culture. The clinicopathological correlation was the reference standard, whereby the presence of infection was deduced from the final diagnosis in each patient's case notes. RESULTS: Our sample included 168 cases (105 positive for infection). GS-FFPE showed a sensitivity of 0.43 (95% confidence interval 0.29, 0.57), a specificity of 0.98 (0.95, 1.01), a positive likelihood ratio of 21.50 (19.76, 23.24), and a negative likelihood ratio of 0.58 (0.41, 0.75). CONCLUSIONS: GS-FFPE has poor sensitivity, and a negative result should not be used as evidence to exclude infection. In contrast, it has excellent specificity and, unless the pretest probability of infection is very low, a positive result would make infection much more likely. The value of the GS-FFPE lies in cases where sterile tissue was not submitted for microbiological studies, or sterile tissue culture was negative, and there is at least a low-to-moderate pretest probability of infection.