Intensive Care Med DOI 10.1007/s00134-008-1015-x ORIGINAL Catharina F. M. Linssen Jan A. Jacobs Jan S. A. G. Schouten Walther N. K. A. van Mook Graham Ramsay Marjolein Drent Influence of antibiotic therapy on the cytological diagnosis of ventilator-associated pneumonia Received: 20 April 2007 Accepted: 6 January 2008 © Springer-Verlag 2008 C. F. M. Linssen (✉) · J. A. Jacobs University Hospital Maastricht, Department of Medical Microbiology, P.O. Box 5800, 6202 AZ Maastricht, The Netherlands e-mail: klin@lmib.azm.nl J. S. A. G. Schouten University Hospital Maastricht, Department of Epidemiology, Maastricht, The Netherlands J. S. A. G. Schouten University Hospital Maastricht, Department of Ophthalmology, Maastricht, The Netherlands W. N. K. A. van Mook · G. Ramsay University Hospital Maastricht, Department of Intensive Care Medicine, Maastricht, The Netherlands M. Drent University Hospital Maastricht, Department of Respiratory Medicine, Maastricht, The Netherlands Abstract Objective: To assess the influence of antibiotics on the value of various cytological parameters, and their combinations, in diagnos- ing ventilator-associated pneumonia (VAP). Design: Prospective study. Setting: The general intensive care unit (17 beds) of the University Hospital Maastricht. Patients: Three hundred and thirty-five episodes of clinically suspected VAP (defined by the clinical and radiological criteria previously described by Bonten et al.) in 282 patients were studied. Inter- ventions: No additional interventions were conducted. Measurements and results: Bronchoalveolar lavage fluid cytology included a total cell count per millilitre, differential cell count and the percentage of infected cells (cells containing phagocytised organ- isms). Antibiotic therapy from 72 h prior to lavage was recorded. Areas under the curve (AUCs) of receiver operating characteristic curves were calculated for various cytological parameters and their combinations, in patients with and without antibiotic therapy. In 126 episodes (37.6%) in 106 patients, VAP was confirmed. There was no difference in AUCs between patients with and without antibiotic therapy for any parameter studied. The most prominent AUCs were (for patient groups with and without antibiotics combined): total cell count, 0.65; percentage poly- morphonuclear neutrophils, 0.71; and percentage infected cells, 0.90. The combination of percentage in- fected cells with any other cyto- logical parameter did not increase the AUC. Conclusion: Antibiotic ther- apy did not influence the predictive value of the percentage infected cells in BALF in diagnosingVAP. Keywords Pneumonia · Broncho- alveolar lavage · Cytology · Anti- biotics Introduction Ventilator-associated pneumonia (VAP) is a common com- plication in intensive care (ICU) patients, represents up to 80% of the nosocomial infections in the ICU [1, 2] and is associated with high mortality rates [1–3]. Bronchoalveolar lavage (BAL) with quantitative cul- ture is currently regarded as a reference method for the microbiologically confirmed diagnosis of VAP [4] since clinical presentation, radiology and laboratory parameters are non-specific. However, results of BAL fluid (BALF) cultures can take up to 48 h. Because microscopic exam- ination of BALF can be achieved within 2 h [5, 6], the diagnostic value of cytological parameters has been ex- plored. Of the previously investigated parameters, such as the total cell count (TCC) [7–9], the percentage of poly- morphonuclear neutrophils (PMNs) [8, 10] and the per- centage of cells containing phagocytised organisms, also known as infected cells or cells with intracellular organ- isms (IC) [6, 11, 12], the last proved to be the most reliable.