Variation of infected cell count in bronchoalveolar lavage and timing of ventilator-associated pneumonia

Objectives. The aim of this study was to evaluate and compare the accuracy of the percentage of infected cells (%IC) in bronchoalveolar lavage (BAL) for ventilator-associated pneumonia (VAP) diagnosis according to its onset from the initiation of mechanical ventilation. Patients. One hundred and eight patients admitted to a surgical ICU were retrospectively included (1999-2001). A total of 171 cases of VAP were diagnosed on clinical, biological, chest X-ray and BAL results (threshold greater than or equal to10(4) cfu/ml). Results. The %IC significantly decreased with the timing of VAP diagnosis: 12.2+/-12.1% for VAP occurring less than 7 days after the initiation of mechanical ventilation, 7.4+/-9.2% for VAP occurring between 7 and 15 days and 4.8+/-6.4% for VAP after 15 days (p=0.0002), despite the same number of elements and proportion of polymorphonuclear neutrophils in BAL. In addition, a relationship between the %IC and the pathogen responsible for VAP was observed for P. aeruginosa [higher for VAP <7 days than for VAP 7-15 days (p=0.01) and VAP >15 days (p=0.006)] and S. aureus [lower for VAP >15 days than VAP 7-15 days (p=0.04) and VAP <7 days (p=0.04)]. Furthermore, the %IC in BAL was lower in patients undergoing antimicrobial therapy than in patients without antibiotics (p=0.04). Three factors were independently associated with the %IC: quantitative culture of BAL (beta=0.42, p<0.0001), ongoing antimicrobial therapy (beta= -0.21, p=0.003) and onset of VAP (beta= -0.17, p=0.01). Conclusions. A relationship between the %IC in BAL, duration of ventilation, quantitative culture of BAL and ongoing antimicrobial therapy has been proved in this study. The %IC for VAP diagnosis may not be accurate in patients with ongoing antibiotics and late onset infections (>7 days).