Value of lower respiratory tract surveillance cultures to predict bacterial pathogens in ventilator-associated pneumonia: systematic review and diagnostic test accuracy meta-analysis.
Intensive Care Med. 2012 Nov 28;
Authors: Brusselaers N, Labeau S, Vogelaers D, Blot S
PURPOSE: In ventilator-associated pneumonia (VAP), early appropriate antimicrobial therapy may be hampered by involvement of multidrug-resistant (MDR) pathogens. METHODS: A systematic review and diagnostic test accuracy meta-analysis were performed to analyse whether lower respiratory tract surveillance cultures accurately predict the causative pathogens of subsequent VAP in adult patients. Selection and assessment of eligibility were performed by three investigators by mutual consideration. Of the 525 studies retrieved, 14 were eligible for inclusion (all in English; published since 1994), accounting for 791 VAP episodes. The following data were collected: study and population characteristics; in- and exclusion criteria; diagnostic criteria for VAP; microbiological workup of surveillance and diagnostic VAP cultures. Sub-analyses were conducted for VAP caused by Staphylococcus aureus, Pseudomonas spp., and Acinetobacter spp., MDR microorganisms, frequency of sampling, and consideration of all versus the most recent surveillance cultures. RESULTS: The meta-analysis showed a high accuracy of surveillance cultures, with pooled sensitivities up to 0.75 and specificities up to 0.92 in culture-positive VAP. The area under the curve (AUC) of the hierarchical summary receiver-operating characteristic curve demonstrates moderate accuracy (AUC: 0.90) in predicting multidrug resistance. A sampling frequency of >2/week (sensitivity 0.79; specificity 0.96) and consideration of only the most recent surveillance culture (sensitivity 0.78; specificity 0.96) are associated with a higher accuracy of prediction. CONCLUSIONS: This study provides evidence for the benefit of surveillance cultures in predicting MDR bacterial pathogens in VAP. However, clinical and statistical heterogeneity, limited samples sizes, and bias remain important limitations of this meta-analysis.
PMID: 23188467 [PubMed - as supplied by publisher]