Influence of Bronchoscopy on the Diagnosis of and Outcomes from Ventilator-Associated Pneumonia

Background: Ventilator-associated pneumonia (VAP) is a common healthcare-associated infection affecting as many as 27% of mechanically ventilated patients. Ventilator-associated pneumonia is an important source of morbidity and mortality in the surgical intensive care unit (SICU). The optimal diagnostic method for VAP has remained controversial and the role of therapeutic bronchoscopy in the clearance of pulmonary secretions with VAP, in essence source control, remains unknown. Our unit utilizes bronchoscopy inconsistently for these purposes and we chose to evaluate its effectiveness in our patient population with the hypothesis that bronchoscopic diagnosis and therapy results in lower mortality rates and faster clinical resolution. Methods: We analyzed retrospectively all patients treated for VAP in a single SICU between September 2003 and December 2011. Patients were divided into groups based upon diagnostic method and receipt of therapeutic bronchoscopy, and were analyzed for differences in time to clinical resolution and mortality. Results: A total of 360 patients were included in the study, including 493 episodes of VAP. The diagnostic bronchoscopy group had statistically higher APACHE II scores (p=0.02) and fewer days in hospital prior to diagnosis (p=0.02) when compared with the non-invasive diagnosis group. Diagnostic bronchoscopy was associated with shorter length of stay and shorter duration of antibiotics whereas receipt of a therapeutic bronchoscopy was associated with the opposite effects by multivariable analysis. Conclusion: Our hypothesis was disproved and our findings are similar to those found in recent publications. This study supports no definitive conclusions, but further consideration of the role of bronchoscopy is urged in both the diagnosis and treatment of VAP. In our population, bronchoscopy for diagnostic or therapeutic purposes in VAP was not associated with better outcomes. However, differences in baseline characteristics suggest a randomized trial may be needed to answer more completely this question.