Gram staining of protected pulmonary specimens in the early diagnosis of ventilator-associated pneumonia

We evaluated prospectively the use of Gram staining of protected pulmonary specimens to allow the early diagnosis of ventilator-associated pneumonia ( VAP), compared with the use of 60 bronchoscopic protected specimen brushes (PSB) and 126 blinded plugged telescopic catheters (PTC) obtained from 134 patients. Gram stains were from Cytospin slides; they were studied for the presence of microorganisms in 10 and 50 fields by two independent observers and classified according to their Gram stain morphology. Quantitative cult ures were performed after serial dilution and plating on appropriate cultur e medium. A final diagnosis of VAP, based on a culture of greater than or e qual to 10(3) c.f.u. ml(-1), was established after 81 (44%) samplings. When 10 fields were analysed, a strong relationship was found between the prese nce of bacteria on Gram staining and the final diagnosis of VAP (for PSB an d PTC respectively: sensitivity 74 acid 81%, specificity 94 and 100%, posit ive predictive value 91 and 100%, negative predictive value 82 and 88%). Th e correlation was less when we compared the morphology of microorganisms ob served on Gram staining with those of bacteria obtained from quantitative c ultures (for PSB and PTC respectively: sensitivity 54 and 69%, specificity 86 and 89%, positive predictive value 72 and 78%, negative predictive value 74 and 84%). Increasing the number of fields read to 50 was associated wit h a slight decrease in specificity and positive predictive value of Gram st aining, but with a small increase in its sensitivity and negative predictiv e value. The results obtained by the two observers were similar to each oth er for both numbers of fields analysed. Gram staining of protected pulmonar y specimens performed on 10 fields predicted the presence of VAP and partia lly identified (using Gram stain morphology) the microorganisms growing at significant concentrations, and could help in the early choice of the treat ment of VAP. Increasing the number of fields read or having the Gram stain analysed by two independent individuals did not improve the results.