Discussion

We found higher resistance among Klebsiella pneumoniae, Escherichia coli, and Enterococcus isolates obtained from a hematology–oncology unit compared to isolates from the hospital-wide antibiogram. We attribute this difference to risk factors for development of MDRO infections, including prolonged neutropenia, extensive hospital stays, and reliance on prophylactic antibiotics. Our results align with those of Rosa et al, ^{Reference Rosa, Simkins and Camargo2} who found that antibiograms for solid-organ transplant units had significantly higher rates of resistance compared to hospital-wide data.

Higher rates of resistance among certain gram-negative organisms suggest a need for heightened stewardship. Even though antibiotics are critical medications, judicious use is necessary to prevent resistance, which, if unchecked, makes classes of antibiotics ineffective for patients who may need them the most. Data suggest that shorter duration of antibiotics for neutropenic fever is not inferior to more prolonged courses. ^{Reference Aguilar-Guisado, Espigado and Martin-Pena8} Stewardship targeting duration of therapy may help prevent or slow development of antibiotic resistance without compromising clinical outcomes.

In addition to increased gram-negative resistance, vancomycin resistance was significantly higher in our unit-specific antibiogram among both E. faecalis and E. faecium. We suspect that this is at least in part due to frequent use of empiric vancomycin in patients with febrile neutropenia, which is often administered outside recommended guidelines. ^{Reference Bartash, Cowman and Szymczak9} Because vancomycin is often started for MRSA coverage, it is important to note that rates of MRSA were similar on both antibiograms. Previous studies have also found low rates of MRSA colonization among hematology–oncology patients. ^{Reference Bartash, Cowman and Szymczak9} One reason for these low rates of colonization may be due to the limited anti-staphylococcal activity of fluoroquinolones, the agents most used for neutropenic prophylaxis. Therefore, although widespread fluoroquinolone use may contribute to increased gram-negative resistance, the same pressure may not apply to Staphylococcus isolates. Given high rates of vancomycin-resistant Enterococcus and >50% methicillin-sensitivity among S. aureus isolates, ongoing stewardship efforts targeting the use of vancomycin on hematology–oncology units is needed.

Interestingly, differences in resistance patterns were not observed for Pseudomonas aeruginosa, the third most commonly isolated gram-negative organism. This finding, similar to that of Kern et al, ^{Reference Kern, Steib-Bauert and de With10} may be due to the nature of acquired resistance mutations for various organisms. Enterobacteriaceae, including K. pneumoniae and E. coli, typically utilize β-lactamases or plasmid-borne cephalosporinases to confer resistance to β-lactam antibiotics, or they develop chromosomal mutations conferring resistance to fluoroquinolones. ^{Reference Ruppé, Woerther and Barbier7} Conversely, Pseudomonas preferentially utilizes specific efflux pumps as a resistance mechanism. Differences in acquisition and mechanism of antibiotic resistance may contribute to the lack of difference noted among our P. aeruginosa isolates, though this may also be explained by higher baseline rates of fluoroquinolone resistance on our hospital-wide antibiogram.

This study had several limitations. The number of unit-specific antibiogram isolates included was small. Based on CSLI recommendations, we were only able to evaluate 6 organisms and additional patterns of resistance might not have been recognized due the low frequencies of other isolates. Additionally, hematologic malignancy patients can be admitted to other hospital units, and their clinical isolates are included in the hospital-wide antibiogram only. Furthermore, the hospital-wide antibiogram included cultures obtained on the hematology–oncology unit, so these data are present in both antibiograms. However, these represent a small proportion of hospital-wide antibiogram isolates, and given the increased resistance seen on the unit-specific antibiogram, inclusion in the hospital-wide antibiogram should have decreased the differences noted between antibiograms.

In summary, hematology–oncology unit-specific antibiograms are important clinical decision tools for empiric treatment of neutropenic fever because increased drug resistance may be seen on these units. Local resistance patterns can help drive antibiotic stewardship efforts in this high-risk population.