Discussion

Culturing the skin using a premoistened sponge had a sensitivity of 92% to detect CRAB carriage. All other screening sites had low sensitivity; even combinations of 2 sites (excluding the skin) reached a maximum of 69% sensitivity. The combination of buccal mucosa and skin was 99% sensitive. Tracheal aspirate, often used to screen for CRAB carriage in ventilated patients,^{Reference Rodríguez-Baño, García and Ramírez6} had a low yield in our study (49%). The yield from skin screening was similar at the 2 study centers, suggesting that it is less influenced by sampling technique or patient characteristics. These results, combined with those of previous studies,^{Reference Doi, Onuoha and Adams-Haduch4,Reference Nutman, Lerner, Schwartz and Carmeli5} suggest that the skin is the main colonization site for CRAB and that it is usually sparse on the skin. Thus, a large surface area is required for sampling using a highly absorbent sponge and enrichment for detection.

Identifying CRAB carriers can direct various infection control activities, including isolation and cohorting of carriers, environmental cleaning, decolonization (eg, chlorhexidine bathing), and removing contact precautions when carriage has resolved. Detection of CRAB carriage can also assist in empiric therapy choice in the event of infection.

The sensitivity of skin screening for CRAB observed in our study (92%) compares favorably with the sensitivity of screening methods commonly used for other multidrug-resistant organisms. For example, culturing the nares for MRSA has up to 90% sensitivity.^{Reference Safdar, Narans, Gordon and Maki7} Screening by rectal swab or stool specimen has 61%–99% sensitivity to detect VRE,^{Reference Faron, Ledeboer and Buchan8} and 76%–100% sensitivity to detect CRE.^{Reference Adler, Navon-Venezia, Moran-Gilad, Marcos, Schwartz and Carmeli9}

An important issue to consider before implementing a screening policy is the pretest probability of screening positive. When the prevalence in the screened population is <10%, <1 positive patient will be missed for every 100 patients screened using a test with 90% sensitivity. When prevalence is higher or when missing a case may have severe consequences, a test with a higher sensitivity should be chosen. Adding a buccal mucosa swab to skin sampling increased sensitivity from 92% to 99%; thus, screening both sites may be preferred in such circumstances.

In a previous study, CRAB active surveillance by rectal swabs and bronchial aspirates using a rapid molecular diagnostic assay that provided results “within a few hours after sample collection,” followed by prompt isolation of carriers, decreased CRAB acquisition by 35% compared to conventional cultures.^{Reference Yamamoto, Hamaguchi and Akeda10} Our methods, although sensitive, are not rapid; they provide a result of “suspected CRAB” in 36 hours and a final result after identification and susceptibility testing, which require an additional 24 hours.

This study has several limitations. Since there is no gold standard for screening, we measured sensitivity at each body site by comparing it to a standard of positivity at any body site. This may be an underestimation, in which case the sensitivities we calculated may be inflated. Also, our study was performed in only 2 centers.

In conclusion, our data support the notion that the skin is the main site of colonization by CRAB, and that culturing the skin using a premoistened sponge as a single site for active surveillance of CRAB is satisfactory. The sensitivity of screening the buccal mucosa, rectum, and tracheal aspirate was low, and a negative screening culture from these sites should not be taken as evidence of CRAB noncarriage.