Validation of an algorithm based on direct examination of specimens in suspected ventilator-associated pneumonia

The management of suspected ventilator-associated pneumonia (VAP) is traditionally based on either a clinical or a "bacteriological" approach. Direct examination of specimens may provide a valuable help to both approaches. The objective was to test an algorithm for the management of suspected pneumonia based on severity criteria and Gram stains of specimens, including early treatment of patients with a positive Gram stain of a protected telescoping catheter specimen and awaiting culture results in patients with either Gram stains of endotracheal aspirate and protected specimen both being negative, or only positive endotracheal aspirate (EA), except in the presence of the severity criteria. Observational study including 76 first episodes of suspected pneumonia in 4 units. We assessed the rates of correctly diagnosed episodes and of those treated empirically (i.e., ultimately confirmed pneumonia), or appropriately not receiving such therapy (non-confirmed pneumonia), based on protected specimen culture results. When adhered to, the algorithm allowed early appropriate management in 80% of patients, including 83% of those with confirmed pneumonia, and 74% of those without confirmed infection. The rate of appropriately managed episodes using this algorithm was significantly higher than that using a strategy based on the modified clinical pulmonary infection score (CPIS) alone (80 vs. 50%, p