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