Discussion

The HOB incidence in the cancer patient population studied was 3.2 per 100 admissions. This incidence is higher than previously reported among patients admitted to acute care hospitals that are not “stand-alone cancer centers” (0.124 per 100 admissions). ^{Reference Yu, Jung and Ai8} This finding suggests that cancer patients are more likely to acquire HOB than the general patient population. A recent large multicenter study reported that patients with myeloproliferative malignancies had up to 6-fold higher risk of HOB events than non-cancer patients, further supporting this hypothesis. ^{Reference Yu, O’Horo, Ai, Jung and Bastow10} The incidence of SOM related HOB incidence was lower than other types of malignancies, which may be due to the increasing use of non-myelosuppressive antineoplastic therapy and immunotherapy. HM patients often have immunodeficiency related to neutropenia and lymphopenia from the disease infiltrating the bone marrow, opposed to SOM that may or may not directly invade the bone marrow and/or compromise neutrophil and cellular components crucial in the immunologic response to pathogens. ^{Reference Cibir, Beer and Kraus11}

Risk factors associated with HOB events in this study included HM, length of stay greater than or equal to 30 days, prior hospitalization within the past 90 days, antibiotic use within the past 90 days, admission into the ICU, development of a CAUTI during admission, and presence of a CVC. In a large multicenter study myeloproliferative cancers and transplant patients were more likely to develop HOB regardless of length of stay. ^{Reference Yu, O’Horo, Ai, Jung and Bastow10} When analyzing HOB cases regardless of underlying conditions, length of stay greater than 30 days was associated with HOB, which aligns with the findings from our study. Patients who stay longer in acute care tend to have more aggregate invasive devices, and with coagulase negative staphylococci as a top HOB pathogen, it is reasonable to assume that repeated or prolonged skin disruption involved in acute care is a factor in HOB events. However, studies have demonstrated that coagulase negative staphylococci and other pathogens listed as CLABSI organisms are in fact present on mucosal surfaces and are likely to be reflective of mucosal barrier injury (MBI) bacteremia. ^{Reference Tamburini, Andermann, Tkachenko, Senchyna, Banaei and Bhatt12} Best practices in judicious use of invasive devices and blood specimen collections will likely need to be revisited to ensure risk mitigation is optimized without compromising quality of care. In addition, enhanced oral care may reduce risk for coagulase negative staph and other oral pathogens in HCT patients. ^{Reference Suwabe, Fuse and Katsura13}

Prior hospitalization at the study center within 90 days was identified as a factor for developing HOB in our population. This association may be due to patients’ recovery from previous hospitalization, ongoing therapy for cancer (particularly HM without HCT or SOM), the development of multi-drug-resistant organisms, or antibiotic exposure which was an additional factor identified to correlate with HOB in this population. Data from CDC’s HAI report identified high-risk antibiotics use as risk for developing sepsis within 90 days of discharge. ^{14,Reference Baggs, Jernigan, Halpin, Epstein, Hatfield and Donald15} Our data similarly suggests recent hospitalization and antibiotic use were risk factors for HOB.

The study population includes patients receiving cancer treatment and HCT who have a higher baseline acuity due to immunosuppression; this elevated risk is further magnified for those requiring ICU admission. Previous literature has reported that that patients admitted to the ICU are more likely to develop HOB due to their acuity and the use of devices similar to our findings. ^{Reference Yu, Ye and Edwards16} A study reported that 7.8% of CAUTI developed HOB and the rates in our patients were also higher in those with HOB. Additionally, the same study found that 3.7% of HOB cases were secondary to a non-CAUTI hospital-onset urinary tract infection (HOUTI). ^{Reference Kelly, Ai, Jung and Yu4} Presences of CVC were associated with higher incidence of HOB, which is expected given that CLABSI is a subset of HOB. However, like prior study findings, non-CLABSI HOB occurred far more often than CLABSI HOB, emphasizing the sources other than these devices. HM patients with HCT are at risk for HOB based on time post HCT (early vs. late). One study reported 17.8% BSI at 30 days post-HCT (with median time to onset of 9 days post-HCT) and 20.5% at day 100 post-HCT with predominant pathogens being coagulase-negative staphylococci and E. coli. ^{Reference Sava, Bättig and Gerull17} Therefore, it is important to develop device-agnostic quality metric in this special population.

Interestingly indwelling urinary catheters were associated with lower incidence of HOBs in our patients compared to the reported literature. In our institution, indwelling urinary catheter use is limited to patients in the ICU, with suspicion of sepsis or shock, undergoing surgical procedures, or experiencing acute urinary retention - each case requiring daily review of necessity. While the CAUTI decision flow chart (Supplemental Figure 1) reflects institutional processes designed to reduce catheter use, low CAUTI rates are largely attributable to the proactive, day-to-day efforts of the infection prevention team, whose surveillance and intervention strategies are critical to success. The lower incidence of HOBs may reflect both high surveillance and restricted use in a specific population. As such, this outcome should be interpreted with caution given the specific population and smaller sample size. Similarly, literature has shown that non-CAUTI HOUTIs are approximately three times more likely to progress to HOB than CAUTIs. ^{Reference Kelly, Ai, Jung and Yu4} However, when a CAUTI did occur in our study, it was twenty-three times more likely to progress into HOB. This suggests that while CAUTIs may occur less frequently, their occurrence represents a particularly high-risk event for bacteremia.

Common HOB organisms in this study such as E. coli and Klebsiella pneumoniae are often associated with urinary tract infections and coagulase-negative Staphylococcus, including Staphylococcus epidermidis, can be associated with CVCs, although data suggests that mucositis and gut microbiome can also be a risk factor for these pathogens. Regarding the latter, Viridians streptococci (11.3% of cases) is a common oral flora that can evolve into a BSI due to mucositis related to antineoplastic therapy. This has been well documented in patients undergoing active chemotherapy and HCT and similarly in our cohort. Alteration of the oral microbiome and cumulative carbapenem use has been shown to be risk factors for Stenotrophomonas maltophilia infections particularly in those with acute myeloid leukemia. ^{Reference Aitken, Sahasrabhojane and Kontoyiannis18} Main risk factors for the development of Pseudomonas aeruginosa blood stream infections in oncological patients are neutropenia, use of corticosteroids, severity of underlying disease, or prior surgery. ^{Reference Tofas, Samarkos and Piperaki19} GI flora is often implicated in MBI related bacteremia and is an NHSN category. This is observed in the context of cytotoxic chemotherapy, acute GVHD of the gut, and leaky gut as seen with cytokine response syndrome in patients with CART cell therapy or immunotherapy. ^{Reference Berg21} Finally, our study highlights a lower HOB incidence for S. maltophilia and P. aeruginosa compared to published data. ^{Reference Karaba, Goodman, Amoah, Cosgrove and Tamma20} Lower rates reflect aggressive antimicrobial stewardship efforts such as an active prospective review and provider feedback of inpatient carbapenem use.

Analyzing the relevant microbiology can therefore inform targeted HOB prevention and care strategies inclusive of aggressive oral and skin hygiene, proper care and choice of invasive device, antimicrobial stewardship, and continued involvement of environmental services. GI source control such as perineal care, wound care and pressure injury ulcer care and prevention are likely partially modifiable risk factors that may mitigate some of the sources for these pathogens. While this study was not designed to assess the preventability of HOB in these clinical conditions, the question of which preventative measures could mitigate HOB in clinical scenarios described above represent areas of further research.