Background: Tracheal aspirate Gram's stains are used to guide antibiotic selection in empiric pneumonia treatment in the surgical intensive care unit (SICU). We questioned whether Gram's stains predict the organism cultured.
Methods: A retrospective review of prospectively collected data.
Results: Gram's stains correlated with the cultured organism in 284 of 543 (52%) SICU cultures and in 226 of 403 (56%) trauma intensive care unit (TICU) cultures. Gram-negative rod (GNR) stains yielded GNR organisms in 182 of 205 (89%) SICU cultures and in 160 of 176 (91%) TICU cultures. Gram-positive coccus (GPC) stains yielded GPC organisms in 75 of 228 (33%) SICU cultures and in 52 of 149 (35%) TICU cultures. Noncorrelates in the GPC group were predominantly GNRs (185 of 250 (74%)).
Conclusion: When the clinical decision has been made that empiric antibiotic coverage is necessary, GNR coverage should be instituted regardless of Gram's stain result. The decision to institute GPC coverage needs to be supported by clinical data other than the Gram's stain.