Can routine surveillance samples from tracheal aspirate predict bacterial flora in cases of ventilator-associated pneumonia?

AIM: The aim of this study was to investigate the hypothesis that periodical sampling of the tracheo-bronchial tree in the absence of clinical suspicion of pneumonia is useful to identify bacteria responsible for subsequent late ventilator associated pneumonia (VAP). This was a retrospective observational human study carried out in two medical-surgical intensive care units of two different hospitals. From January 1999 to December 2000, 559 patients, who received invasive respiratory support for more than 48 hours, were screened. METHODS: Tracheal aspiration (TA) was performed once or twice weekly in all mechanically ventilated patients. The microbiological findings from TA surveillance cultures done in the eight days before suspicion of VAP were compared to those isolated from the positive diagnostic samples done for late onset VAP (after more than four days of mechanical ventilation). The sensitivity, specificity, and positive/negative predictive values of the ability of the surveillance sample to anticipate the VAP pathogen were calculated. RESULTS: Among the microorganisms isolated from TA, 68% were retrieved from diagnostic samples. All VAP pathogens previously isolated were from 43% of the TA samples. If TA was collected 2-4 days before the clinical diagnosis of VAP, pathogens were detected in 58% of samples. In contrast, only 27% were collected more than four days earlier (P<0.05). The positive predictive values for Pseudomonas aeruginosa and methicillin resistant Staphylococcus aureus from routine TA samples were 92% and 90%, respectively. The negative predictive values for Pseudomonas aeruginosa and methicillin resistant Staphylococcus aureus from routine TA samples were 75% and 89%, respectively. CONCLUSIONS: TA cultures in ventilated patients may help to predict pathogens. Early detection may lead to earlier treatment in long term ventilated patients when VAP is suspected.