The Performance of Gram-Staining in Tailoring the Empirical Antibiotic Choice in Operated Diabetic Foot Infections

Current international guidelines recommend choosing an empirical antibiotic regimen for treating diabetic foot infections (DFI) based largely on the clinical severity of infection and the local microbiological epidemiology. This may lead to selecting unnecessarily broad spectrum initial empiric antibiotic therapy. Using data from our hospital in a large Swiss city, we retrospectively analyzed the performance of the Gram-stained smears of predominantly deep surgical DFI specimens processed by our microbiology laboratory in predicting the microorganism grown on standard cultures. We excluded episodes with paucibacillary stain results, which we interpret as contamination. Among 1,235 operated moderates or severe DFIs, Gram-stained smear was reported in 321 (26%) of cases, and showed bacteria in 172 episodes (54%) of these. Overall, among Gram-stain results with organism seen, the sensitivity, specificity, positive and negative predictive values of the Gram-stain smear when compared with the cultures was 56%, 93%, 97%, and 38%, respectively. The corresponding statistical values specifically for Gram-negative bacteria were 61%, 97%, 50%, and 82%. We conclude that the results of routine Gram-stain smears of deep intraoperative DFI specimens generally lack sufficient sensitivity, and was only useful to reasonably exclude a DFI caused predominantly by gram-negative bacteria. For Gram-stained smears results to be useful for guiding antibiotic stewardship, we need prospective trials to assess its value in different types of DFIs.