In this surveillance study, we focused on two questions which are cardinal in limiting the spread of MDR A. baumannii
: which body site should be cultured in order to detect carriage of MDR A. baumannii
, and what is the duration of carriage? We found that culturing a single body site has very low sensitivity, not higher than 30%, and that even when multiple sites are sampled the sensitivity of detecting carriers of MDR A. baumannii
reaches only 55%. Sampling multiple body sites is time-consuming and costly, and it is not suitable for routine use by clinicians and clinical laboratories. The low sensitivity of single-site surveillance for MDR A. baumannii
is in contrast to the much higher sensitivities of single-site screening for other MDR pathogens, e.g., rectal cultures for the detection of carriers of vancomycin-resistant enterococci and nasal cultures to detect carriers of methicillin-resistant Staphylococcus aureus
). We also found that MDR A. baumannii
may be carried for long durations, up to 42 months, and that prolonged carriage affects at least 17% of patients with previous clinical isolations of MDR A. baumannii
. This proportion of long-term carriers is likely an underestimate due to the limited sensitivity of surveillance in the detection of carriers (55%) and implies that prolonged carriage of MDR A. baumannii
may affect 30% of patients with remote clinical isolations.
A combination of several factors may explain the low sensitivity of surveillance cultures. First, it was assumed that all patients from whom MDR A. baumannii was isolated from clinical cultures within the last 10 days were carriers, and this may not have been true in all cases, i.e., certain patients may have had MDR A. baumannii at the infection site only. This is likely the explanation in very few cases, since both surveillance sites and clinically relevant sites (draining wounds and endotracheal aspirates from intubated patients) were sampled. Second, patients may have received an appropriate antimicrobial therapy against MDR A. baumannii, which may have eradicated the pathogen prior to our sampling. However, among our cohort, only two had received such a regimen, as one patient received cefepime (for which the in vitro results have a questionable significance if an extended-spectrum-β-lactamase-producing organism is present), and the second patient received, for only 1 day prior to culture, colistin. Therefore, we believe that this factor did not considerably bias our results. Third, the sampling and microbiological methods used may not be sensitive enough to detect MDR A. baumannii bacteria, particularly if they are present at the sampled body sites in low concentrations. Although an enrichment method was used after its superiority to direct culturing onto selective media was confirmed, this method may still not be sensitive enough. Regarding the surveillance skin samples, for example, we can only postulate that another method which samples a skin surface area larger than that sampled by swabbing would increase the yield of the surveillance cultures. Fourth, MDR A. baumannii may occupy different body sites in different patients.
One might expect that the pharynx would be the best site to sample for carriage of MDR A. baumannii
for a number of reasons: Acinetobacter
spp. were reported to be common commensals in the human pharynx, pneumonia is the most common clinical syndrome of A. baumannii
infections, and the bacterium is known for its ability to rapidly colonize tracheotomies (2
). In fact, we found that the pharynx had a low sensitivity (23%) as a surveillance site. The skin was also suggested to be appropriate site for the surveillance of MDR A. baumannii
). In our cohort of patients with remote clinical isolations, four of five long-term carriers were identified by skin sampling. However, the skin had the lowest sensitivity as a surveillance site for patients with recent clinical isolation (13.5%), though as we previously mentioned, the appropriate methodology for obtaining skin samples is yet not well defined.
Isolating MDR A. baumannii
from hospitalized patients depends on external ecological variables and risk factors related to the patients themselves (5
). Several previous reports have discussed the risk factors associated with the development of MDR A. baumannii
infections in hospitalized patients (2
). As far as we know, this is the first report that investigates the risk factors associated with prolonged MDR A. baumannii
carriage. Two of the risk factors identified, a bedridden state and a disoriented state, were both reported in the past as being associated with MDR A. baumannii
infections in hospitalized patients (2
). Small sample sizes limit the generalizability of these results and do not allow for meaningful multivariate analysis. In addition, due to the low sensitivity of surveillance and the possible resultant misclassification of MDR A. baumannii
carriers as noncarriers, the identification of these risk factors should be interpreted cautiously.
Another study limitation was that the previous MDR A. baumannii
strains of the remote carriers were not available for genotyping, and therefore we can only assume that remote carriers remained carriers of the same clone. However, when we genotyped isolates that were colonizing concurrently different body sites, we saw that a given clone can be isolated from multiple sites and cause different clinical syndromes. In addition, the clones identified were all familiar, having been previously associated with MDR A. baumannii
outbreaks at TASMC (1
), and therefore it is reasonable to assume that the long-term carriers acquired the strains in the nosocomial setting, continued to harbor them for several years, and probably spread them in their local outpatient environment as well.
In conclusion, our study demonstrates that the current methodology to detect MDR A. baumannii carriage is suboptimal and that persistent carriage of MDR A. baumannii occurs in a substantial proportion of patients. Improved methods of surveillance are necessary, and long-term contact precautions for MDR A. baumannii carriers should be considered.