Surveillance Cultures in Healthcare-Associated Pneumonia
Surveillance Cultures in Healthcare-Associated Pneumonia
Methods and results of studies concerning the use of surveillance cultures in VAP management are summarized in Table 1. One systematic review and 24 studies were included. The systematic review concerning the value of surveillance cultures of ETA included 791 episodes of clinically suspected or confirmed VAP derived from 14 studies. The review calculated a pooled sensitivity between 72 and 84%, and a pooled specificity between 90 and 98% for surveillance cultures of ETA in predicting the bacterial pathogen in VAP. However, in many studies incorporated in the review as well as in many of the 24 studies presented in Table 2, some methodological concerns arise.
First, most studies concerned retrospective analysis of VAP episodes, and randomized controlled trials were lacking. The former may reveal higher predictive values for surveillance cultures compared with the value of surveillance cultures by taking all obtained surveillance cultures into account (the latter being a more correct method). Second, no worldwide accepted gold standard was available for diagnosing VAP and studies used different diagnostic approaches, while not always distinguishing colonization from infection. When ETA is used for VAP diagnosis, it is not correct to compare micro-organisms identified by ETA with VAP-causing micro-organisms, the latter also being based on ETA results. This so-called incorporation bias, which overestimates diagnostic accuracy, was most likely present in a significant number of studies. In some studies, incorporation bias may be present, because diagnostic methods were not clearly described. Third, studies were difficult to compare because of the heterogeneity in study population (postcardiac surgery, medical ICU), MDR prevalence, surveillance culture sampling sites and frequency, as well as the previously discussed method for VAP diagnosis. Furthermore, cutoff for positive ETA surveillance cultures varied. Some included surveillance cultures of ETA samples with at least 10 cfu/ml, whereas others included surveillance cultures of ETA samples with less than 10 cfu/ml. Other studies compared individual micro-organisms identified in surveillance cultures, thereby artificially increasing the specificity by increasing the numbers of 'true negatives'. Finally, some studies presented data from cases in the same hospital and same period, suggesting publication bias.
After taking into account these concerns, some studies suggest that more appropriate antibiotic treatment is given when guided by available surveillance cultures compared with ATS guidelines. or a hypothetical antibiotic treatment model. However, the methodology used herein was perhaps less suited to demonstrate such correlation, and MDR presence may have been caused by other risk factors such as previous antibiotic treatment and prolonged mechanical ventilation (potential confounders). Whereas improved survival is suggested by one study, others demonstrate equal mortality and no study was found to be sufficiently powered to demonstrate any mortality differences.
Surveillance cultures of gastric samples appeared not useful for antibiotic guidance in VAP management. In one study, 33% of all VAP causative pathogens were previously detected by nasal swab, rectal swab or urine culture. Sensitivity increased by 13–18% in two studies when the results of all performed surveillance cultures were taken into account. However, the expected negative effect on positive predictive value and specificity of this strategy (taking into account all surveillance cultures) was not provided.
Overall, sensitivity of previously available surveillance cultures for VAP seems moderate to even high, and therefore could more appropriately guide antibiotic treatment, especially in the case of P. aeruginosa and MDR. However, a decrease in infection attributable mortality when surveillance cultures were performed was so far not demonstrated. The positive predictive value of surveillance cultures seems low and, in our opinion, surveillance culture results should therefore always be used together with clinical criteria before the initiation of antibiotics.
Surveillance Cultures for Management of Ventilator-Associated Pneumonia
Methods and results of studies concerning the use of surveillance cultures in VAP management are summarized in Table 1. One systematic review and 24 studies were included. The systematic review concerning the value of surveillance cultures of ETA included 791 episodes of clinically suspected or confirmed VAP derived from 14 studies. The review calculated a pooled sensitivity between 72 and 84%, and a pooled specificity between 90 and 98% for surveillance cultures of ETA in predicting the bacterial pathogen in VAP. However, in many studies incorporated in the review as well as in many of the 24 studies presented in Table 2, some methodological concerns arise.
First, most studies concerned retrospective analysis of VAP episodes, and randomized controlled trials were lacking. The former may reveal higher predictive values for surveillance cultures compared with the value of surveillance cultures by taking all obtained surveillance cultures into account (the latter being a more correct method). Second, no worldwide accepted gold standard was available for diagnosing VAP and studies used different diagnostic approaches, while not always distinguishing colonization from infection. When ETA is used for VAP diagnosis, it is not correct to compare micro-organisms identified by ETA with VAP-causing micro-organisms, the latter also being based on ETA results. This so-called incorporation bias, which overestimates diagnostic accuracy, was most likely present in a significant number of studies. In some studies, incorporation bias may be present, because diagnostic methods were not clearly described. Third, studies were difficult to compare because of the heterogeneity in study population (postcardiac surgery, medical ICU), MDR prevalence, surveillance culture sampling sites and frequency, as well as the previously discussed method for VAP diagnosis. Furthermore, cutoff for positive ETA surveillance cultures varied. Some included surveillance cultures of ETA samples with at least 10 cfu/ml, whereas others included surveillance cultures of ETA samples with less than 10 cfu/ml. Other studies compared individual micro-organisms identified in surveillance cultures, thereby artificially increasing the specificity by increasing the numbers of 'true negatives'. Finally, some studies presented data from cases in the same hospital and same period, suggesting publication bias.
After taking into account these concerns, some studies suggest that more appropriate antibiotic treatment is given when guided by available surveillance cultures compared with ATS guidelines. or a hypothetical antibiotic treatment model. However, the methodology used herein was perhaps less suited to demonstrate such correlation, and MDR presence may have been caused by other risk factors such as previous antibiotic treatment and prolonged mechanical ventilation (potential confounders). Whereas improved survival is suggested by one study, others demonstrate equal mortality and no study was found to be sufficiently powered to demonstrate any mortality differences.
Surveillance cultures of gastric samples appeared not useful for antibiotic guidance in VAP management. In one study, 33% of all VAP causative pathogens were previously detected by nasal swab, rectal swab or urine culture. Sensitivity increased by 13–18% in two studies when the results of all performed surveillance cultures were taken into account. However, the expected negative effect on positive predictive value and specificity of this strategy (taking into account all surveillance cultures) was not provided.
Overall, sensitivity of previously available surveillance cultures for VAP seems moderate to even high, and therefore could more appropriately guide antibiotic treatment, especially in the case of P. aeruginosa and MDR. However, a decrease in infection attributable mortality when surveillance cultures were performed was so far not demonstrated. The positive predictive value of surveillance cultures seems low and, in our opinion, surveillance culture results should therefore always be used together with clinical criteria before the initiation of antibiotics.