Use of a rapid detection test for extended-spectrum beta-lactamase from direct blood culture: an antimicrobial stewardship tool in a tertiary hospital in Spain

Abstract

Introduction

Bacteremia caused by Gram-negative bacilli places a substantial burden on healthcare systems, mainly due to antibiotic resistance and delays in administering appropriate antimicrobial treatment (AT). The aim of this study was to describe the implementation of a rapid detection test (RDT) for CTX-M Extended Spectrum Beta-Lactamase-producing Enterobacterales (ESBL-PE) bacteremia as a tool for Antimicrobial Stewardship (AMS) in a tertiary hospital in Spain.

Material and methods

A cross-sectional study was conducted on blood culture (BC) samples from adult patients (≥18 years) admitted to a tertiary hospital in Spain (January 2021-February 2024). BCs with confirmed Enterobacterales identification were included. An RDT was used to detect CTX-M ESBL from direct BC. The results were reported to the AMS team. Data from electronic medical records and our laboratory information system were analyzed to explore the utility of implementing an RDT as an AMS tool.

Results

A total of 250 BCs from 250 patients were included. Empiric antimicrobial treatment (EAT) had not been prescribed in 41/250 (16.4%) patients, but was appropriately initiated in 33/250 (13.2%) after notification of the RDT results. Among those already receiving EAT (209/250, 83.6%), inappropriate and appropriate actions in AT were observed in 18/250 (7.2%) and 191/250 (76.4%) patients, respectively. By the time routine AST results were available, 241 (96.4%) patients had received appropriate treatment.

Conclusions

This study demonstrated the real-world application of an RDT to detect CTX-M ESBL directly from BC in a tertiary hospital. Early reporting of CTX-M ESBL status in Enterobacterales bacteremia enabled physicians and AMS teams to optimize AT.

Introduction

Bacteremia caused by Gram-negative bacilli is common in both ambulatory and hospitalized patients [1] and the burden of these infections on healthcare systems is substantial [2], with mortality rates of up to 20% [3]. Major contributing factors include the presence of antimicrobial-resistant strains and the time required to administer appropriate antimicrobial treatment (AT) [4–6]. In a recent meta-analysis, the initiation of inappropriate empirical antimicrobial treatment (EAT) for Enterobacterales bacteremia was associated with higher mortality, especially in patients with a mixed Enterobacterales and Extended Spectrum Beta-Lactamase (ESBL)-producing Enterobacterales (ESBL-PE) bacteremia [7]. Moreover, another meta-analysis showed that appropriate EAT in ESBL-PE bacteremia was protective against mortality [8]. Therefore, early detection of antimicrobial resistance mechanisms is important for the initiation of appropriate treatment or the modification of ongoing treatment.

The widespread adoption of proteomic methods in clinical laboratories has led to a reduction in the time required to identify microorganisms from direct blood cultures (BC). However, there is also an urgent need to develop strategies for the rapid optimization of AT. Among possible interventions, the introduction of rapid diagnostic tests (RDTs) to assess bacterial resistance into clinical practice could reduce inappropriate AT choices [9]. RDT-guided therapies improve antibiotic use by facilitating timely adjustment of AT, thereby reducing unnecessary prescribing of broad-spectrum AT [10]. In addition, rapid identification of resistant pathogens allows for prompt infection control measures, limiting the spread of multidrug-resistant organisms and improving patient safety in healthcare settings [11].

In 2022, third-generation cephalosporin-resistant Escherichia coli and Klebsiella pneumoniae accounted for 23.5% of bacteremia cases in Europe [12]. From 2018 to 2022, the incidence of bacteremia due to third-generation cephalosporin-resistant E. coli and K. pneumoniae in Spain was 14.0% and 26.3%, respectively [12]. ESBLs are one of the most common resistance mechanisms, and microorganisms harboring ESBLs colonize over 1.5 million people [13]. Several risk factors related to ESBL-PE bacteremia have been identified, including the use of a permanent urinary catheter and AT in the previous 3 months [14]. In Spain, the prevalence of CTX-M enzymes has evolved, with current reports including groups 14, 15, 24, and 27 [15]. The aim of this study was to describe the practical implementation and initial findings of using an RDT to detect CTX-M ESBL directly from BC as an Antimicrobial Stewardship (AMS) tool in a real clinical setting of a tertiary hospital in Spain.

Methods

Study design. This study was conducted at the Microbiology Department of a 1,308-bed tertiary hospital in Madrid, Spain. BC samples flagged as positive were collected prospectively from November 2021 to February 2024. The microbiological criteria for BC samples to be included in the study were a morphology compatible with Gram-negative bacilli observed by Gram staining and the identification of a microorganism from the order Enterobacterales by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS). Only monomicrobial bacteremias were included. The clinical criteria for inclusion in the study were adult patients (18 years and older) admitted to the Emergency department or hospitalized and treated with common groups of antibiotics used for empirical treatment in our hospital, including amoxicillin/clavulanic acid, piperacillin/tazobactam, third- or fourth-generation cephalosporins, carbapenems, fluoroquinolones, and aminoglycosides. For patients with a confirmed beta-lactam allergy, the prescription of fluoroquinolones or aminoglycosides was considered appropriate. Patients receiving alternative antibiotics were excluded to maintain consistency in the comparison of empirical therapy approaches based on local guidelines. Patients who had not yet received antimicrobial treatment at the time of the positive blood culture were also included. Detection of CTX-M ESBL from direct BC was performed regardless of previous ESBL carrier status. Although multiple BC samples were flagged as positive in a patient’s bacteremic episode, only the first BC was used for CTX-M ESBL testing. Subsequent blood cultures were not included in this analysis but were used for routine clinical follow-up to monitor ongoing bacteremia and therapeutic response. Antimicrobial susceptibility testing was also performed and concordance with the CTX-M ESBL result by RDT was verified.

Sample collection, transport, and processing. BC samples were collected in the Emergency department or during hospitalization according to in-hospital, standardized procedures [16]. After collection, the BC bottles were immediately transported to the Microbiology laboratory, where samples are processed 24 hours a day, 7 days a week. There are two work shifts: from 8 am to 3 pm, blood cultures are processed by laboratory technicians, and from 3 pm to 8 am, blood cultures are processed by Microbiology residents.

The samples were then loaded into the automated BD BACTEC^™ FX unit (Beckton Dickinson, Madrid, Spain) from 2021 to mid-2023. From then on, the samples were incubated into the automated BACT/ALERT® VIRTUO® unit (bioMérieux, Marcy l’Etoile, France). A 5-day incubation protocol was set in both instruments. If a BC was flagged as positive, Gram staining and protein extraction were performed immediately for subsequent identification by MALDI-TOF MS (Biotyper® Bruker® Daltonics GmbH & Co. KG, Bremen, Germany) [17]. For protein extraction, a rapid method involving Triton X-100 detergent solution and formic acid was used, as previously described and optimized in our laboratory to improve identification accuracy. If Gram-negative bacilli were observed on Gram staining and the MALDI-TOF MS analysis from direct BC identified a microorganism belonging to the order Enterobacterales, the eligibility criteria for the detection of CTX-M ESBL from direct BC were then applied. Simultaneously, aerobic BC bottles were cultured at 37°C on blood agar and chocolate agar (in a 5% CO₂ atmosphere) for 24 hours, and anaerobic BC bottles were cultured at 37°C on blood agar (in a 5% CO₂ atmosphere) and Brucella agar (in an anaerobic atmosphere) for 2 days.

VITEK® 2 System: routine antimicrobial susceptibility testing (AST). A two-step centrifugation process was used to prepare a purified bacterial suspension from the direct BC sample, which was later adjusted to a 0.7-1 McFarland standard, as previously described by Romero-Gómez et al. [18]. Antimicrobial susceptibility testing (AST) was then performed using the VITEK® 2 System (bioMérieux, France) according to the manufacturer’s instructions [19]. AST-425 cards were used for Gram-negative bacteria AST, which includes detection of ESBL. AST results are expressed as the minimum inhibitory concentration (MIC) in milligrams per liter, and a preliminary report is available in 6 to 8 hours [19]. In our Microbiology laboratory, a set of BC samples flagged as positive with Gramnegative bacilli of the order Enterobacterales is sufficient criteria to perform AST. AST results are validated in 12-24 hours and results are reported to the Antimicrobial Stewardship (AMS) team.

NG-Test® CTX-M MULTI: rapid diagnostic test (RDT). NG-Test® CTX-M MULTI (NG Biotech Laboratories, Guipry, France) is a qualitative lateral flow immunochromatography assay (LFIA) designed to detect CTX-M ESBL (groups 1, 2, 8, 9, and 25) from grown colonies. A procedure for CTX-M ESBL testing using NG-Test® CTX-M MULTI from direct BC was previously described and evaluated by Cendejas-Bueno et al. [20]. Briefly, one hundred microliters of the 0.7-1 McFarland standard from the previous step (routine AST) is added to the sample well. After a 15-minute incubation at room temperature, the results are interpreted as follows: CTX-M ESBL-positive, a red line appears on both the control and test regions; CTX-M ESBL-negative, a red line appears only on the control region. Results are available to both the attending physician and the AMS team within 1 to 1.5 hours of the BC being flagged as positive.

Antimicrobial Stewardship (AMS) team. The AMS team at our hospital consists of clinical microbiologists, pharmacologists, and infectious disease specialists. The infectious disease specialists are responsible for recommending the initiation or modification in AT to the attending physicians based on clinical data and microbiological information. Daily face-to-face meetings are held in the Microbiology Department from Monday to Friday at 10 am to share relevant microbiological results after sample processing during the morning (8 am to 10 am) and provide updates on final AST results. An emergency number is available for real-time communication of results after this time. This line is also available for night shifts and weekends, as blood cultures are continuously processed as they are flagged as positive.

Exploring the potential utility of RDTs in AMS. A review of electronic medical records and the laboratory information system was used to collect demographic, clinical, and laboratory data. Antimicrobial therapy was reviewed three times: (1) after the CTX-M ESBL testing in the laboratory, (2) again 10-12 hours after notification of the CTX-M ESBL result to both the attending physician and the AMS team, and (3) a third time when AST results were already available and the AMS team had made further recommendations for AT, within the first 12-24 hours after the BC was flagged as positive (Figure 1). EAT for patients in the Emergency department or hospitalization was determined by physicians based on local guidelines and in-hospital susceptibility to the major microorganisms associated with bacteremia (Enterobacterales). Definitions for antimicrobial escalation or de-escalation in AT were obtained from a previously published consensus [21]. In addition to these definitions, appropriate actions in AT also included the continuation of appropriate AT, or discontinuation of antibiotics for Gram-positive coverage. Optimal antibiotic treatment after the notification of the CTX-M ESBL result was defined as the prescription of amoxicillin, amoxicillin/clavulanate or piperacillin/tazobactam for anaerobic coverage, and cefotaxime or ceftriaxone for CTX-M ESBL-negative Enterobacterales bacteremia, cefepime for ampC-overproducing Enterobacterales bacteremia and carbapenem for CTX-M ESBL-positive Enterobacterales bacteremia [22]. Time to appropriate AT was defined as the time from BC collection to appropriate de-escalation of AT or time to appropriate escalation of AT. Mortality was also recorded at the third medical record review.

Figure 1.

Implementation of a rapid detection test (RDT) for extended-spectrum beta-lactamase (ESBL) into the routine blood culture (BC) processing workflow.

*Proteomic identification by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS).

1 AT, antimicrobial treatment.

Statistical analysis. Categorical variables were reported as numbers and percentages, and continuous variables were summarized as medians with interquartile ranges (IQR). The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of the RDT (NG-Test® CTX-M MULTI) were calculated with 95% confidence intervals (95% CI) using Stata version 18.0 (StataCorp, College Station, TX, USA).

Results

Participant characteristics. A total of 2,155 participants with a positive flagged BC were screened for eligibility during the study period. 21/2,155 (0.97%) participants were excluded due to prescription of alternative antibiotics, as they did not align with the predefined empirical treatment criteria. Among the remaining participants, a positive Gram stain for Gram-negative bacilli followed by proteomic identification of Enterobacterales was found in 250 participants. The median age of the participants was 74 years (IQR, 58.5 to 86) and 51.6% were women (Table 1). The most common coexisting comorbidities were cardiovascular (42.8%) and oncological (33.6%) disease. Risk factors for ESBL infection were present in 40.4% of participants, including the use of a permanent urinary catheter, AT within the prior 3 months, and hospitalization within the prior 3 months. The genitourinary system (42%) was the most common source of bacteremia, followed by the digestive system (32.8%) and the respiratory system (11.6%). 13/250 (5.2%) presented with sepsis and 6/250 (2.4%) with septic shock at the time of blood culture collection.

Table 1.

Baseline characteristics of the study population (N=250).

	Total
Age, median (IQR), years	74 (58.5-86)
Female, no. (%)	129 (51.6)
Comorbidities, no. (%)
Cardiovascular disease	129 (42.8)
Oncological disease	98 (39.2)
Renal disease	27 (10.0)
Neurological disease	22 (7.2)
Urological disease	21 (6.4)
Respiratory disease	17 (5.6)
Gastrointestinal disease	5 (2.0)
Risk factors for ESBL infection, no. (%)
Hospitalization within prior 3 months	43 (17.2)
Antimicrobial treatment within prior 3 months	38 (15.2)
Use of permanent urinary catheter	16 (6.4)
Previous ESBL colonization	4 (1.6)
Source of bacteremia, no. (%)
Genitourinary	105 (42.0)
Digestive	82 (32.8)
Respiratory	29 (11.6)
Skin and soft tissue	14 (5.6)
Central venous catheter	12 (4.8)
Unknown	8 (3.2)

IQR, interquartile range; ESBL, extended-spectrum beta-lactamase.

Microbiological features related to Gram-negative bacteremia. Most BC were extracted at the Emergency department (66%) and from peripheral venipuncture (88.4%). The median time for BC to be flagged as positive after incubation was 9.6 hours (IQR, 8.2 to 12.1). E. coli and K. pneumoniae were the most common microorganisms detected from the order Enterobacterales, accounting for 71.2% and 23.6%, respectively (Table 2).

Table 2.

Results of the rapid diagnostic test (RDT) for the detection of CTX-M ESBL, and the antimicrobial susceptibility test (AST) in Enterobacterales detected in positive flagged blood cultures (N=250).

Microorganism, n (%)	Total n=250	RDT CTX-M ESBL positive n=48	AST CTX-M ESBL positive n=47	Other resistance mechanisms* n=10
Escherichia coli	178 (71.2)	32 (66.7)	31 (65.9)	3 (30)**
Klebsiella pneumoniae	59 (23.6)	16 (33.3)	16 (34.1)	7 (70)***
Klebsiella oxytoca	6 (2.4)
Proteus mirabilis	3 (1.2)
Klebsiella aerogenes	2 (0.8)
Citrobacter koseri	1 (0.4)
Enterobacter cloacae	1 (0.4)

RDT, rapid diagnostic test; AST, antimicrobial susceptibility test; ESBL, extended-spectrum beta-lactamase.

^*Other resistance mechanisms were detected using phenotypic methods with the AmpC Confirm Kit (Rosco Diagnostica, Taastrup, Denmark) and a multiplex PCR (Seegene Allplex^™ Entero-DR Assay, Seoul, South Korea) to identify eight resistance genes, including OXA-48, VIM, KPC, NDM, IMP, vanA and vanB.

^**1 pAmpC, 1 NDM+CTX-M ESBL, 1 OXA-48.

^***1 OXA-48, 1 VIM, 4 OXA-48+CTM-M ESBL, 1 KPC+CTX-M ESBL.

The sensitivity, specificity, positive and negative predictive values with 95% CI of NG-Test® CTX-M MULTI were 100% (92.45-100), 99.5% (97.3-99.9), 97.9% (86.9-99.7) and 100% (98.2-100). A positive result for CTX-M ESBL was detected in 18.8% of the microorganisms by both RDT and AST (Table 2). One microorganism classified as CTX-M ESBL positive by the RDT was not subsequently confirmed by the AST and no other resistance mechanism was detected; this was considered as a false positive. Resistance mechanisms, other than CTX-M ESBL, were detected by the AST in 10% of the microorganisms. These included the AmpC beta-lactamase and the carbapenemases OXA-48, KPC, NDM and VIM. The presence of these resistance mechanisms did not interfere with the RDT performance, even in CTX-M ESBL-positive strains.

Antimicrobial stewardship: RDT and AST in Enterobacterales bacteremia In 41/250 (16.4%) patients, EAT had not yet been prescribed at the time of the first review of the electronic medical records. Of them, AT was initiated inappropriately in 8/41 (19.5%) patients and appropriately in 33/41 (80.5%) patients after notification of the microorganism detected on BC and the RDT result, according to the second review of the electronic medical records (Figure 2). In the remaining patients, EAT was already initiated (209/250, 83.6%). Inappropriate and appropriate actions in AT were observed in 18/209 (8.6%) and 191/209 (91.4%) of them, respectively, after reporting the microorganism detected on BC and the RDT result. One patient with E. coli bacteremia had a false positive result with the RDT. This patient had a confirmed beta-lactam allergy and, although an inaccurate result was reported, continuation of appropriate AT with fluoroquinolones was recorded throughout the duration of AT. Overall, 241/250 (96.4%) patients had received appropriate treatment when the results of the routine AST were available and the third review of the electronic medical records was performed. A second antibiotic for Gram-positive coverage was prescribed in 18/250 (7.2%) and was discontinued in 7/250 (2.8%) and 11/250 (4.4%) patients after the RDT and AST results, respectively. In 7/250 (2.8%) patients, there were situations in which it was not possible to record interventions in AMS (e.g., death or transfer to another hospital within 24 hours of admission).

Figure 2.

Changes in antimicrobial therapy after the notification of the CTX-M Extended Spectrum Beta-Lactamase (CTX-M ESBL) result by rapid detection test (RDT) and the antimicrobial susceptibility test (AST) in Enterobacterales bacteremia.

After notification of a CTX-M ESBL-positive Enterobacterales bacteremia in patients treated with EAT (n=34), a higher proportion of appropriate actions in AT occurred with third-generation cephalosporins, with 9/34 (26.5%) escalations of AT, and with carbapenems, with 11/34 (32.4%) continuations of AT (Table 3). And, after notification of a CTX-M ESBL-negative Enterobacterales bacteremia (n=157), a higher proportion of appropriate actions in AT occurred with third-generation cephalosporins, with 75/157 (47.8%) continuations of AT, and with piperacillin/tazobactam, with 40/157 (25.5%) continuations of AT (Table 3).

Table 3.

CTX-M Extended Spectrum Beta-Lactamase (CTX-M ESBL) detection from direct blood culture by rapid detection test (RDT): results by antibiotic group.

	No antimicrobial treatment				Empirical antimicrobial treatment
	BLEE +		BLEE −		BLEE +		BLEE −
	Appropriate n=7	Non-appropriate n=1	Appropriate n=26	Non-appropriate n=7	Appropriate n=34	Non-appropriate n=6	Appropriate n=157	Non-appropriate n=12
Third generation cephalosporins, n=100
Initiation, no. (%)		1 (100)	11 (68.8)
Continuation, no. (%)						2 (33.3)	75 (47.8)
Escalation, no. (%)					9 (26.5)
De-escalation, no. (%)
Other
Delay in initiation, no. (%)				2 (11.8)
Piperacillin/tazobactam, n=65
Initiation, no. (%)			10 (58.8)
Continuation, no. (%)						4 (66.7)	40 (25.5)
Escalation, no. (%)					5 (14.7)			5 (41.7)
Other
Delay in initiation, no. (%)				1 (5.9)
Amoxicillin/clavulanic acid, n=30
Initiation, no. (%)			5 (31.2)
Continuation, no. (%)							23 (14.6)
Escalation, no. (%)					3 (8.8)
Carbapenems, n=32
Initiation, no. (%)	6 (85.7)			2 (11.8)
Continuation, no. (%)					10 (29.4)			7 (58.3)
De-escalation, no. (%)							6 (3.8)
Fluoroquinolones, n=12
Continuation, no. (%)							1 (0.6)
Escalation, no. (%)					5 (14.7)		5 (3.2)
De-escalation, no. (%)							1 (0.6)
Fourth generation cephalosporins, n=7
Initiation, no. (%)				2 (11.8)
Continuation, no. (%)							4 (2.5)
Escalation, no. (%)					1 (2.9)
Aminoglycosides, n=4
Initiation, no. (%)	1 (14.3)
Continuation, no. (%)					1 (2.9)
De-escalation, no. (%)							2 (1.3)

Inappropriate continuation of third-generation cephalosporins and piperacillin/tazobactam was observed in 2/100 (2%) and 4/65 (6.2%) patients after the notification of a CTX-M ESBL-positive Enterobacterales bacteremia, respectively. Inappropriate escalation from piperacillin/tazobactam to carbapenems, and inappropriate continuation of AT with carbapenems was observed in 5/65 (7.7%) and 7/32 (21.9%) patients, respectively, after the notification of a CTX-ESBL-negative Enterobacterales bacteremia.

After the notification of the RDTs results, the median time to initiation of AT was 5.5 hours (IQR, 4-9.3), and the median time to appropriate escalation and appropriate de-escalation was 4 hours (IQR, 2-5.2) and 16 hours (IQR, 3.2-5.1), respectively. The median duration of AT was 7 days (IQR, 3-10.7). Deaths occurred in 13/250 (5.3%) patients after optimization of AT with the RDT result, with a median time of 10 days (IQR, 3-16). Of them, only one patient had a CTX-M ESBL-positive Enterobacterales bacteremia and escalation from amoxicillin/clavulanic acid to meropenem was made in 5.1 hours after notification of the results. In those with a CTX-M ESBL-negative Enterobacterales bacteremia, 3/13 (23.1%) initiated appropriate AT and 8/13 (61.5%) continued appropriate EAT. Optimization of AT (de-escalation from carbapenems to third-generation cephalosporins) was performed in 2/13 (15.4%) patients with the RDT results, in 3.9 hours. Secondary de-escalation was performed in 5/13 (38.5%) after the AST results.

Discussion

In this study, a high proportion of appropriate actions in AT, including continuation, escalation or de-escalation, was observed in patients with Enterobacterales bacteremia following the implementation of an RDT for the detection of CTX-M ESBL from direct BC. A total of 89.6% of patients were prescribed appropriate AT using the RDT, either in patients without prescribed AT at the time of result notification or in patients with EAT.

Culture-based methods for subsequent identification and susceptibility testing of cultured microorganisms typically require an additional 24-48 hours after a blood culture is flagged as positive [23]. The use of RDTs such as immunoassays or molecular tests has reduced the time required for these microbiological procedures. Clinical benefit may also be achieved by prompt initiation of targeted AT or modification of EAT, particularly in severe medical conditions such as bacteremia or sepsis. Therefore, the utility of these tests in several types of samples, including grown colonies from blood cultures and directly from urine samples, has recently been investigated for use in conjunction with coordinated in-hospital AMS programs [24,25].

Prospective audit and feedback over a 48-72-hour period is a priority intervention in AMS [26,27]. This is an opportunity to perform a medical chart review with the aim of optimizing antibiotic selection, dose, route, frequency, and duration after microbiological data are available [28]. As a result, the implementation of RDTs focused on microorganism identification (e.g., MALDI-TOF) and detection of resistance mechanisms (e.g., ESBL] represents a major advance in AMS interventions [29]. RDTs facilitate decision-making at different stages of AT, including initiation, continuation, escalation, de-escalation, or discontinuation of AT [9].

A recent meta-analysis found a high sensitivity and specificity of several RDTs when used to detect ESBL from direct BC [30]. In our study, the diagnostic performance of the RDT in terms of sensitivity, specificity, positive and negative diagnostic values was similar to these results. Furthermore, LFIA RDTs offer additional advantages due to their low cost, ease of use and lack of need for expert personnel [31]. A study on the impact of LFIA RDTs for the detection of ESBL from direct BC was recently conducted in Italy [32]. A total of 199 BC samples with E. coli also identified by MALDI-TOF were included. Although no mention was made regarding the implementation of RDTs in a pre-existing AMS team, the results were similar to ours. However, a meta-analysis of 6 randomized controlled trials including 1,638 participants showed with a low level of certainty that RDTs for the detection of antimicrobial resistance in bacteremia had no advantage over conventional tests, even for outcomes such as mortality or time to appropriate AT [33].

In our study, a subset of patients had no prescribed EAT at the time of the first medical chart review. In these cases, where bloodstream infection was clinically suspected but not yet confirmed microbiologically, physicians may have prioritized further diagnostic evaluations or awaited confirmatory results before initiating treatment. This underscores the importance of integrating rapid diagnostic tools into clinical workflows to provide timely, actionable data that can support earlier therapeutic decisions. For the majority of these patients, appropriate actions in AT were initiated promptly, reflecting efforts to minimize delays and optimize care. Inappropriate actions in AT were also observed despite the availability of microbiological information, including RDT results. These were likely influenced by the complexity of the clinical presentations and associated comorbidities, which may have led physicians to prioritize perceived safety over microbiological guidance. In certain instances, delays in the initiation of AT or the use of inappropriate regimens may also reflect the challenges of managing critically ill patients with multiple competing clinical priorities. However, most of these initial inappropriate actions were subsequently corrected after the availability of AST results, emphasizing the value of combining RDT with routine AST to optimize therapy.

For patients with prescribed EAT, the most appropriate actions in AT (de-escalation and escalation) were performed within the first 24 hours after the notification of the bacteremia and the RDT results. A high proportion of escalations were made earlier than de-escalations. Several reasons could explain this finding. The initial response to bacteremia often involves escalation to broad-spectrum antibiotics to ensure coverage against potential pathogens while awaiting more detailed diagnostic information. This proactive approach aims to stabilize the patient and control the spread of infection. Additionally, the urgency to treat severe infections aggressively from the onset typically leads to early escalation, while de-escalation decisions are made more cautiously and typically after confirming that the patient is stable, and the specific pathogen is sensitive to narrower-spectrum antibiotics. Moreover, institutional protocols and clinical guidelines may also prioritize escalation to ensure immediate patient safety, with de-escalation occurring later as part of a reassessment process.

A low proportion of deaths was observed in this study, which is lower than mortality rates reported in similar hospital settings [34,35]. Most of the patients who died had a CTX-M ESBL-negative Enterobacterales bacteremia, with appropriate AT decisions more often guided by AST results than by the RDT data. The associated comorbidities in these patients made changes in AT with the RDT result less safely performed by physicians. This suggests that while RDTs provide rapid diagnostic information, the presence of other complex medical conditions requires a cautious approach to modifying AT. Future studies should evaluate whether the implementation of rapid diagnostic tests could influence mortality rates in similar contexts.

The main strength of this study is that it provides real-world experience of using an RDT for antimicrobial resistance in a tertiary hospital where an AMS team is already established. By combining an ESBL-specific RDT with MALDI-TOF MS, microbiological information could be provided earlier to the AMS team to guide appropriate AT or to support continuation of AT. In addition, the diagnostic performance of the RDT was high and proved to be well adapted to the epidemiology of CTX-M ESBL in our hospital. In consequence, the results supported its implementation in our current BC workflow routine. There are also some limitations. First, electronic chart review was used to collect patient’s data, and information bias is possible. Some clinical data regarding patient status or severity of infection may be evolving or under investigation at the time of electronic review, and decisions regarding AT interventions may not have been adequately recorded. Second, a control group was not available and comparisons with a pre-intervention group were not made, although the diagnostic performance of the RDT offers an important clinical value for a rapid optimization of AT. Third, the implementation of this RDT may be difficult in a low-resource setting or in healthcare facilities with a different workflow (e.g., no MALDI-TOF available). However, in settings where the CTX-M ESBL is considered to be prevalent, Gram staining could guide the use of the RDT efficiently [9]. Fourth, in our study, a favorable epidemiological context (a high CTX-M ESBL prevalence) enabled the use of negative CTX-M ESBL results by RDT to perform changes in AT (e.g., de-escalation of Gram-negative antibiotic coverage, including carbapenems). Nonetheless, in other epidemiological contexts (a low CTX-M ESBL prevalence), only a positive CTX-M ESBL result by RDT would be considered a safe approach to modify AT (e.g., escalation). Otherwise, other changes in AT, such as de-escalation, would be better applied when the AST results are available, as there is evidence of better clinical outcomes in Gram-negative bacteremia when appropriate AT is optimized. Fifth, while this study focused primarily on the initial utility of RDTs on antimicrobial therapy, future research could examine long-term outcomes such as recurrence of infection, development of antimicrobial resistance, or readmission rates. Monitoring these data would provide further evidence of the effectiveness of RDTs in improving patient care beyond the acute hospital stay.

Conclusions

This study presented the results of a real-world experience of implementing an RDT for the detection of CTX-M ESBL from direct BC into a 24-hour turnaround workflow that included the identification of microorganisms, also from direct BC. These RDTs allowed preliminary information on CTX-M ESBL-producing Enterobacterales bacteremia to be reported on the same day the BC was flagged as positive. These techniques have the potential to support clinical decision-making in the optimisation of AT by both clinicians and the AMS team.

Funding

None to declare.

Conflict of interest

Authors declare no conflict of interest.