Impact of Adding a Rapid PCR-Based Blood Culture Identification Panel to the Antimicrobial Stewardship Program of Patients with Febrile Neutropenia in a Peruvian Referral Hospital

The addition of Biofire® FilmArray® Blood Culture Identification panel 2 (BCID2) to the antimicrobial stewardship program (ASP) could improve outcomes in bloodstream infections (BSI) of patients with febrile neutropenia (FN). A pre- and post-quasi-experimental single-center study was conducted at a reference hospital in Peru. Three groups were considered: patients with BSI before ASP intervention (control group), patients with BSI after ASP intervention (group 1), and patients with BSI after ASP intervention plus BCID2 PCR Panel implementation (group 2). Overall, 93 patients were identified (32 control, 30 group 1, 31 group 2). The median time to effective therapy was significantly shorter in group 2 compared to group 1 and control group, respectively (3.75 vs. 10 h, p = 0.004; 3.75 vs. 19 h, p < 0.001). No significant differences in terms of relapse of bacteremia, in-hospital mortality (all cause), and 30-day-all-cause hospital readmission between the three study periods were found. The appropriateness of empirical antimicrobial use, adding or change, and the following de-escalation or discontinuation was significant when the two intervention periods were compared with the control group (p < 0.001). In addition to the lack of local studies documenting the microbiological profile of FN episodes, adding syndromic panels-based testing could allow for the consolidation of ASP strategies.


Introduction
Despite the improvement on preventative measures and treatment of febrile neutropenia (FN), it is still one of the most common complications in patients undergoing cancer chemotherapy, with an estimated in-hospital mortality rate of 10% [1]. Bacterial infections are common complications in patients with FN, and the epidemiology of bloodstream infection (BSI) episodes has changed in recent years [2]. Antibiotic resistance is a global threat [3], and special gram-negative bacilli have emerged as an important problem in different clinical settings, including hematological patients [4]. Moreover, the dynamics of antimicrobial resistance are influenced by local epidemiology, intensive chemotherapy, the use of invasive devices, and antimicrobial prophylaxis [5].

Results
Overall, 93 patients with positive blood cultures who met the inclusion criteria were identified during the three study periods, with a total of 123 isolates (41 in control group, 37 in period 1, and 45 in period 2). The baseline characteristics of the patients were similar between the intervention groups and control group (before ASP implementation), except for a higher proportion of women and skin and soft tissue infections as a presumed source of bacteremia in period 2.
A higher Charlson comorbidity index was observed in group 2. A lower proportion of patients who received antibacterial prophylaxis was also observed in both intervention periods (Table 1). In control group, 11 patients received prophylaxis with ciprofloxacin and 9 with trimethoprim/sulfamethoxazole. In the intervention periods there was no prescription of quinolones as prophylaxis. During period 2, all 9/9 (100%) patients received prophylaxis with trimethoprim/sulfamethoxazole.
Microbiology. BSI episodes were distributed as follows: 75.6% gram-negative, 22% gram-positive bacteria, and 2.4% for Candida spp. (Figure 1). No significant differences between the three study periods were found. Overall, 23 episodes of polymicrobial bacteremia were identified during the three study periods (24.7% of patients). The three most frequently isolated gram-negative species were E. coli, Klebsiella pneumoniae, and Pseudomonas aeruginosa during the three follow-up periods.
The median time to blood culture positivity was significantly shorter during the intervention period with BCID2 panel compared to the two previous periods (11.14 vs. 18 h; respectively) (p < 0.001). The median time to organism identification from the time of blood culture collection was significantly shorter in the intervention period with BCID2 panel (19.5 h) compared to the ASP implementation period (48 h) and control group (55 h) (p < 0.001). There were no differences in time to in vitro susceptibility results between the three study periods (Table 2).  Table 3. Three cases of discrepancy in polymicrobial cultures were detected. In patient #3, Klebsiella pneumoniae was identified by both methods, however the BCID2 panel additionally identified Acinetobacter calcoaceticus-baumannii complex, Streptococcus spp. and E. coli; this involves a change in antimicrobial therapy to colistin plus meropenem and ampicillin/sulbactam. The antimicrobial therapy in patient #11 was vancomycin + piperacillin/tazobactam, even though Streptococcus could not be isolated by conventional methods. Patient #31 was changed from piperacillin/tazobactam to meropenem after the detection of the CTX-M gene, however Proteus could not be isolated, unlike E. coli and Serratia marcescens. bacteremia were identified during the three study periods (24.7% of patients). The three most frequently isolated gram-negative species were E. coli, Klebsiella pneumoniae, and Pseudomonas aeruginosa during the three follow-up periods. The median time to blood culture positivity was significantly shorter during the intervention period with BCID2 panel compared to the two previous periods (11.14 vs. 18 h; respectively) (p < 0.001). The median time to organism identification from the time of blood culture collection was significantly shorter in the intervention period with BCID2 panel (19.5 h) compared to the ASP implementation period (48 h) and control group (55 h) (p < 0.001). There were no differences in time to in vitro susceptibility results between the three study periods (Table 2).    In patient #4, the BCID2 panel did not detect the presence of ESBL in Klebsiella oxytoca, unlike what was reported by the VITEK-2 system. In patient #25, the mec A/C gene was not detected by the BCID2 panel either.
Acinetobacter lwoffii and Pseudomonas putida were identified in two cases by conventional culture, and both are not included in the BCID2 panel. In patient #27, the BCID2 panel identified Staphylococcus epidermidis (mec A/C gene); however, the initial preliminary examination by the microbiology laboratory indicated yeast growth, with isolation and definitive report of Candida kefyr. Due to this discrepancy, the patient received vancomycin and caspofungin with microbiological clearance of fungemia in control cultures at 48 h.
Regarding resistance profiles, an increase in cases of carbapenemase-producing Enterobacteriaceae (CPE) was observed for period 2 with the detection of the KPC gene in four isolates and OXA-48 gene in one isolate. During the two previous periods, no cases of CPE were detected. Likewise, in period 2, Pseudomonas aeruginosa isolates had a pan-susceptible phenotype compared to control group, where rates of MDR and extensively drug-resistant (XDR) phenotypes were 42.9% and 57.1%, respectively. After ASP intervention and BCID2, one case of Enterococcus faecium with vancomycin resistance (van A/van B gene) and linezolid resistance was detected (report from VITEK-2 automated system) ( Table 4).
Antimicrobial utilization. The appropriateness of empirical antimicrobial use and adding or changing and the following de-escalation or discontinuation was significant when the two intervention periods were compared with the control group (p < 0.001) ( Table 5). However, no statistically significant differences between the two intervention periods were observed. The appropriateness use of empiric therapy in intervention period 2 was 100%.
Treatment and Clinical outcomes. The median time to effective therapy was significantly shorter after ASP intervention and BCID2 compared to ASP intervention and control group, respectively (3.75 vs. 10 h, p = 0.004; 3.75 vs. 19 h, p < 0.001). No significant differences in terms of relapse of bacteremia, in-hospital mortality (all cause), 30-day-all-cause hospital readmission, and length of stay (LOS) following the first positive blood culture between the three study periods were found. However, hospital LOS was significantly shorter between ASP intervention and the BCID2 period vs. control group (28 vs. 33 days, p = 0.016) ( Table 6). Overall mortality was 100% (n = 5) on patients who presented bacteremia due to carbapenemase-producing Klebsiella pneumoniae, with 3/5 (60%) presenting persistent bacteremia. Three patients with KPC-producing Klebsiella pneumoniae and one patient with OXA-48-producing Klebsiella pneumoniae bacteremia received colistin-based combination regimens.

Discussion
This study presents the first evaluation, to our knowledge, of the impact of a Rapid PCR-Based Blood Culture Identification Panel on the activities of an ASP in Peru. The study focuses on the appropriateness of antimicrobial use in different steps for a population of immunocompromised patients. In this study, we found that the addition of the BCID2 panel to ASP improves the time to effective therapy compared to the period before ASP intervention and after ASP intervention alone. These findings were similar to those reported by MacVane SH et al., where a small proportion of patients were neutropenic [24].
The clinical impact on cancer patients from rapid microbiological diagnostic tests, for example BioFire ® FilmArray ® sepsis panel (BCID) or MALDITOF (Matrix Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry), have limited data [25,26]. A quasi-experimental study reported by Buss et al. [27] analyzed the impact of BCID in treating bacteremia in cancer patients over a 4-year period. Three groups were compared, a pre-BCID period (n = 52), after BCID and before the ASP intervention (post-BCID, n = 43), and after the BCID and ASP intervention (post-ASP, n = 35). There was no significant difference in time to appropriate therapy (before BCID: Moreover, a retrospective study by Rosa et al. of 95 patients with hematologic malignancies or bone marrow transplantation showed no benefits over time for appropriate therapy using multiplex PCR blood panel identification alone or in conjunction with the ASP intervention [28]. There are limitations that may explain the non-benefit of BCID combined with ASP strategies, e.g., the general lack of understanding of these technologies among health workers. Therefore, it would be necessary to educate and explore the advantages and limitations of this approach in the onco-hematological patient population [29]. To minimize the limitation of not being familiar with this technology, we developed antibiotic treatment algorithms based on results from the BCID2 panel. To address potential uncertainties generated by reports, particularly with regard to resistance genes, we provided training and immediate feedback to both the hematology service and infection prevention and control unit.
No outcome differences on mortality at 30 days, readmission, or relapse of bacteremia were observed, even though the Charlson comorbidity index was higher in period 2. These results differ from those reported by Rosa, R.G. et al. [18], and Madran, B. et al. [19], where lower mortality rates are reported after the implementation of the ASP. Our study found a shorter hospital LOS in period 2 compared to the control group. While this finding suggests the potential for cost savings, further investigation is required to confirm the impact of ASP on this crucial metric. Specifically, longer follow-up periods and more robust data are needed to assess the impact of ASP on hospital LOS. Nevertheless, the initial reduction in LOS observed in our study provides important insight, especially for low-and middle-income countries where cost savings are critical [30].
Gram-negative bacteria were predominant as BSI agents during the three evaluated periods. This information is relevant since there are few reports of the epidemiology of BSI episodes during FN episodes in Peru. Hinojosa-Andía et al. reported a predominance of gram-negative bacteremia in two years of follow-up in another social security hospital, as ESBL-producing Klebsiella pneumoniae is related to mortality in patients with acute leukemia who received reinduction chemotherapy [13].
Beyond the agents that are not included in the BCID2 panel, a good performance was evident [23], especially in CPE and ESBL Enterobacteriaceae. In one patient with Klebsiella oxytoca isolate, the BCID2 panel did not identify the ESBL gene. This limitation occurs in countries with a high prevalence of ESBL genes such as Peru, whereas several types circulate (e.g., TEM, SHV) [31]. In the current era of rapid diagnostic technologies, combining different diagnostic methods to identify antimicrobial resistance mechanisms can lead to discordant results when compared to conventional microbiology laboratory data. Additionally, in the case of polymicrobial results, the use and interpretation of these tools require rationalization to enable individualized clinical decision-making. To address these challenges, it is essential to optimize the interpretation of antimicrobial susceptibility testing. Our study emphasizes the need for a case-by-case approach to the interpretation of discordant results, taking into consideration individualized clinical scenarios.
Furthermore, the use of BCID2 panel allowed us to have rapid identification of resistance genes in period 2, highlighting the rise of CPE. This period coincides with the increase of bacterial resistance during the COVID-19 pandemic [32] and highlights the importance of adjusting the empiric approach for antimicrobial treatment [14,33]. Empirical therapy protocols in our center are based on the initial use of piperacillin/tazobactam or meropenem in a context of high ESBL prevalence. However, these changes in the epidemiology of BSI episodes make it necessary to collect more data by center and country within the region in order to update the mentioned protocols [4].
The overall mortality in patients with episodes of bacteremia due to CPE was 100%, similar to the report by Jaiswal et al. in India [34]. Inadequate empirical antimicrobial therapy in high-risk neutropenic patients with gram-negative bacteremia has been reported to be associated with increased mortality [35,36]. We observed high rates of appropriateness in empirical therapy according to the initial consensus algorithm on the two intervention periods, but appropriateness de-escalation or continuation decreased in period 2 since addressing emerging resistance mechanisms were not contemplated. Additionally, three cases of relapse of CPE bacteremia were documented in patients with colistin-based regimens. Therefore, it is necessary to evaluate the real effectiveness of colistin in these scenarios.
The targeted treatment of infections caused by CPE on the hematology ward was difficult because our center does not count with alternative drugs to colistin or tigecycline, which are recommended by some guidelines [33,37]. For example, ceftazidime/avibactam is recommended as an alternative for severe infections due to KPC-producing or OXA-48 producing Enterobacteriaceae for the management of FN in patients with hematological malignancies, despite the lack of well-designed comparative studies [33,38]. The ATLAS global surveillance program from 2017 to 2019 collected information from the Latin American region and reported in vitro susceptibility to ceftazidime/avibactam and colistin in 99.4% and 74.9% in non-metallo-SS-lactamase CPE (n = 358), respectively [39]. In a scenario like Peru, where there is also an emergence of colistin-resistant Klebsiella pneumoniae strains [40], it is necessary to consider the inclusion of new SS-lactam-SS-Lactamase Inhibitor combinations or novel agents (e.g., cefiderocol), considering the increase in bone marrow transplant patients (as seen in period 2) [41]. Based on the emergence of CPE and the increase in hematopoietic transplant recipients in our hospital, it is necessary to reinforce the active surveillance system for multidrug-resistant organisms (MDROs) (pre-transplant and pre-engraftment), and through longitudinal studies, the impact of new combinations of antimicrobials administered early in colonized patients could be evaluated.
Almenara Hospital has had an operating ASP since 2017, with initial objectives being met after its implementation, such as the reduction in the antimicrobial consumption rates of imipenem and vancomycin [20]. These antibiotics exert selection pressure for the appearance of MDROs. In intervention period 2, no episodes of MDR or XDR Pseudomonas aeruginosa bacteremia were detected. This could be a trend associated with the effect of the ASP intervention. Medium-term objectives such as changes in the resistance profile of some germs such as Pseudomonas aeruginosa were documented in other areas of our center (ICU) in 2018 [42]. In light of the emergence of multidrug-resistant Enterobacteriaceae and evolving resistance patterns of Pseudomonas aeruginosa observed in our hospital, there is a need to explore non-carbapenem options for treating this group of pathogens. Ceftolozane/tazobactam has been recommended by some guidelines as targeted therapy against Difficult-To-Treat Resistance (DTR) Pseudomonas aeruginosa. However, further prospective evaluation is necessary to determine its efficacy in managing febrile neutropenia [33,38].
These data confirm the enormous gap in conventional microbiological diagnosis in our institution and a molecular microbiological approach, for which we consider that resuming efforts should be prioritized to continue with ASP strategies. The hematology ward was one of the first to join the continuous effort of the strategy through the development of algorithms for the empirical treatment of FN, the discussion of antifungal use strategies, dosing of vancomycin trough levels, among other interventions to the group of patients with hematological malignancies (including autologous transplants).
Our study has some limitations. First, it is likely that patients in the intervention period may have more isolates or different epidemiology based on previous ASP interventions, which reinforced microbiological aspects (timely collection of blood cultures, sample quality, among others) and the prescription of empirical antimicrobials based on a treatment algorithm. This could influence the better selection and de-escalation of antimicrobials; however, this does not translate into a lower selection of MDROs. Additionally, this is a single-center study, with a small sample size and, in period 2, is limited by the budget allocated to the study. The results and outcomes should be interpreted with caution. However, to reduce selection bias, we were rigorous in identifying cases in the control and period 1. Nevertheless, there are potential unmeasured confounders such as limitations in hospitalizations due to outbreaks (the service only has 11 beds), variations in hospital stay periods, and the contamination ratio of blood cultures, among others. Second, many of the ASP strategies, as well as in other local institutions and worldwide, have been affected during the COVID-19 pandemic. ASP members were distributed by contingency issues. Additionally there is a decreased laboratory capacity due to staff desertion or redistribution of functions [43]. Third, it was not possible to determine if the decrease in indications for antibacterial prophylaxis with quinolones contributed to CPE increase in the intervention period 2. Although there are reviews that indicate that their use has no impact on the survival of patients with FN, it does not influence the selection pressure by MDROs [44], for which it is necessary to re-evaluate alternative strategies in contexts of the high prevalence of ESBL. Fourth, despite the local initiatives of our hospital in order to consolidate an ASP, there are still no regulations for all the social security centers in Peru and reports of clinical or impact indicators of these programs are still very scarce. We believe that political barriers and the availability of human resources assigned exclusively to these programs limit interventions. Nevertheless, our study possesses several strengths. It is the first study to describe the impact of an ASP intervention over a 5-year period in Peru, demonstrating an increase in the appropriateness rate of antimicrobial use at various time points when compared to the control period. Additionally, the BCID2 intervention period significantly shortened the time to effective therapy ( Figure 2). Furthermore, the initial reduction in hospital LOS observed in our study highlights the potential of ASP to reduce healthcare costs and lays the foundation for future interventions aimed at strengthening these programs. Through further research in this area, we can enhance our understanding of the impact of ASP on hospital LOS and improve patient outcomes.

Materials and Methods
Study design. This was a pre-post-quasi experimental single-center study conducted at the Guillermo Almenara Irigoyen National Hospital in Lima, Peru. This is a reference center for the insured population in Peru with a capacity of 1116 beds and 23,113 annual discharges. The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Guillermo Almenara Irigoyen National Hospital (protocol code 27-2021).
The study population included all patients older than 18 years old who were admitted to the adult hematology ward (11 beds), with a clinical-laboratory diagnosis of FN and bloodstream infection (BSI) from October 2015 to October 2022. Patients with identification of bacteremia prior to admission to the adult hematology ward, previous bacteremia by the same microorganism, and isolates that were considered contaminants by the ASP team were excluded.
The following three groups were considered: patients with BSI from October 2015 to October 2018 (before ASP intervention-control group), patients with BSI from October 2018 to October 2021 (after ASP intervention and without BCID2 PCR Panel-group 1), and patients with BSI from October 2021 to October 2022 (after ASP intervention plus BCID2 PCR Panel implementation-group 2). Convenience sampling was performed in group 2 for the budget designated for this study. All consecutive cases that met the inclusion criteria for period 2 were considered.
Procedures. Evaluation of patients for inclusion in the study was performed by consulting the database of positive blood cultures from the microbiology laboratory. Only the first positive blood culture was considered for each patient. Once the patients were identified, the electronic hospital record or clinical history was used to collect the variables of interest. Patients with incomplete records of clinical, microbiological, or treatment outcomes were not considered.
Data in the intervention period (period 2) were obtained upon discharge of the pa-

Materials and Methods
Study design. This was a pre-post-quasi experimental single-center study conducted at the Guillermo Almenara Irigoyen National Hospital in Lima, Peru. This is a reference center for the insured population in Peru with a capacity of 1116 beds and 23,113 annual discharges. The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Guillermo Almenara Irigoyen National Hospital (protocol code . The study population included all patients older than 18 years old who were admitted to the adult hematology ward (11 beds), with a clinical-laboratory diagnosis of FN and bloodstream infection (BSI) from October 2015 to October 2022. Patients with identification of bacteremia prior to admission to the adult hematology ward, previous bacteremia by the same microorganism, and isolates that were considered contaminants by the ASP team were excluded.
The following three groups were considered: patients with BSI from October 2015 to October 2018 (before ASP intervention-control group), patients with BSI from October 2018 to October 2021 (after ASP intervention and without BCID2 PCR Panel-group 1), and patients with BSI from October 2021 to October 2022 (after ASP intervention plus BCID2 PCR Panel implementation-group 2). Convenience sampling was performed in group 2 for the budget designated for this study. All consecutive cases that met the inclusion criteria for period 2 were considered.
Procedures. Evaluation of patients for inclusion in the study was performed by consulting the database of positive blood cultures from the microbiology laboratory. Only the first positive blood culture was considered for each patient. Once the patients were identified, the electronic hospital record or clinical history was used to collect the variables of interest. Patients with incomplete records of clinical, microbiological, or treatment outcomes were not considered.
Data in the intervention period (period 2) were obtained upon discharge of the patient and the information obtained was compiled in a database stored in the Excel v.2019 program. To obtain the total number of isolates in period 2, the isolates obtained by both BCID2 and the microbiology laboratory were combined. Informed consent was not used, since the interventions for the microbiological diagnostic approach of FN are part of the standard protocols in the institution.
Definitions: Febrile neutropenia (FN) is defined as an oral temperature of >38.3 • C or two consecutive readings of >38.0 • C for 2 h and an absolute neutrophil count (ANC) of <0.5 × 10 9 /L, or they were expected to fall below 0.5 × 10 9 /L [1].
Bloodstream infection (BSI), demonstrated by positive blood culture, included primary bacteremia/fungemia and secondary bacteremia/fungemia observed while they were neutropenic in the same episode of FN. Contamination is considered to be the isolation of common species in a single blood culture bottle (e.g., coagulase-negative Staphylococcus).
The Multinational Association of Supportive Care in Cancer (MASCC) risk index score was applied to determine the risk of serious complications during FN [45]; episodes were classified as high risk if the score was <21 points and as low risk if the score was ≥21 points.
The appropriateness of antimicrobial use was assessed in three steps: 1. Empirical antimicrobial therapy. 2. Addition or change of antimicrobials. 3. Appropriate continuation, de-escalation, or discontinuation.
Intervention strategy: For ASP implementation in 2018, a focus group methodology was used, which included professionals from the infectious disease (ID) unit, microbiology department, infection preventionist, hematologists, clinical pharmacists, and nurses. Based on prioritized axes, functions and strategies by health professional groups were established. The antimicrobial management protocols for FN episodes were revised and agreed upon, as well as antimicrobial and antifungal prophylaxis (infectious disease (ID) unit and hematologists). An ID member of the ASP evaluated the appropriateness of the antimicrobial prescriptions (the rate of appropriateness in starting, adding, switching, de-escalating, and discontinuing antimicrobials was analyzed). Infection preventionist measured compliance with transmission-based isolation precautions, handwashing behavior, and coordinated educational activities to optimize cleaning and disinfection measures. The surveillance of MDR organisms and telephone reporting of results during the intervention period (group 2) were the responsibility of the microbiology staff. Clinical pharmacist developed antimicrobial infusion protocols and the antimicrobial consumption rates. Nurses optimized infusion practices and received training for blood culture collection.
The ASP strategy was both prospective and restrictive antimicrobial audit (for highcost antimicrobials), however the positive culture review strategy was applied in intervention group 2. Recommendations for changing antimicrobial therapy, if necessary, could be made from 08:00 a.m. to 06:00 p.m., Monday-Friday. Results outside these hours could be performed the next business day. During the intervention period with BCID2, the microbiology laboratory staff communicated the results electronically (WhatsApp telephone network) to the ASP team. The results were published in the medical records after verbal notification.
Identification and susceptibility testing were performed using conventional phenotypic methods and a VITEK-2 system. The FilmArray panel was granted a trial period at the hospital during 2020-2022. Additional molecular diagnostic methods were carried out in the molecular biology laboratory of the Universidad Peruana de Ciencias Aplicadas (UPC) if necessary. At the time of BCID2 implementation, the presence of carbapenemases was detected using the modified carbapenem inactivation method (mCIM), and the immunochromatographic test (K-set CORIS Bio-Concept RESIST-4 O.K.N.V). The FilmArray BCDI2 panel (BioFire Diagnostics, Salt Lake City, UT, USA.) was processed from positive blood cultures as of 23  The discrepancies between BCID2 and conventional culture and phenotypic antimicrobial susceptibility test results were described. Additionally, the resistance profiles of the BSI episodes were compared between the three periods.
Outcomes. The primary outcome was the comparison of times to effective therapy between the three groups. The time to effective therapy was defined as the period between the index blood culture collection and the receipt of the initial dose of an antimicrobial shown to exhibit activity against the patient-specific organism based on in vitro susceptibility results, with intermediate results considered ineffective [24].
Clinical endpoints were compared between groups and included in-hospital mortality (all cause), 30-day all-cause readmission, hospital length of stay (LOS), and LOS following first positive blood culture. All-cause mortality was defined as death resulting from any cause at the end of hospitalization. A relapse of BSI was defined as the reoccurrence of the same organism in a blood culture within 30 days after the end of treatment.
Statistical analysis. Statistical comparisons were made between intervention groups and the control group using the Kruskal-Wallis or the Mann-Whitney test for continuous variables, as appropriate. The chi2 test or Fisher's exact test was obtained for dichotomous variables. Statistical analysis was performed using Stata SE 15.0 software for Windows (College Station, TX, USA). Graphs were created with GraphPad Prism 9.0.0. A p value ≤ 0.05 was considered statistically significant.

Conclusions
In addition to the lack of local studies documenting the microbiological profile of FN episodes in Peru, we considered that adding syndromic panels-based testing will allow for the consolidation of ASP strategies. Likewise, this study could generate evidence for the subsequent implementation of these tests on an institutional level.  Informed Consent Statement: Informed consent was not necessary for participation in this study since there was no direct contact with the participants during the control period and period 1. Medical data were reviewed from secondary sources, such as the microbiology database, during these periods. During the intervention period, data were obtained upon discharge of the patient. Consistent with routine diagnostic procedures for febrile neutropenia, we did not request any additional samples that required specific informed consent. We took measures to protect participant privacy by ensuring that no identifiable information was included in the study data. In addition, we kept a separate record (securely and with access only to the research team) that linked participant codes to their identification, allowing for analysis of clinical information in conjunction with data obtained upon discharge.

Data Availability Statement:
The data supporting the reported results are available from the corresponding author upon reasonable request.