Getting rapid diagnostic test data into the appropriate hands by leveraging pharmacy staff and a clinical surveillance platform: a case study from a US community hospital

Abstract

Objectives

To outline the procedural implementation and optimization of rapid diagnostic test (RDT) results for bloodstream infections (BSIs) and to evaluate the combination of RDTs with real-time antimicrobial stewardship team (AST) support plus clinical surveillance platform (CSP) software on time to appropriate therapy in BSIs at a single health system.

Methods

Blood culture reporting and communication were reported for four time periods: (i) a pre-BCID [BioFire^® FilmArray^® Blood Culture Identification (BCID) Panel] implementation period that consisted of literature review and blood culture notification procedure revision; (ii) a BCID implementation period that consisted of BCID implementation, real-time results notification via CSP, and creation of a treatment algorithm; (iii) a post-BCID implementation period; and (iv) a BCID2 implementation period. Time to appropriate therapy metrics was reported for the BCID2 time period.

Results

The mean time from BCID2 result to administration of effective antibiotics was 1.2 h (range 0–7.9 h) and time to optimal therapy was 7.6 h (range 0–113.8 h) during the BCID2 Panel implementation period. When comparing time to optimal antibiotic administration among patients growing ceftriaxone-resistant Enterobacterales, the BCID2 Panel group (mean 2.8 h) was significantly faster than the post-BCID Panel group (17.7 h; P = 0.0041).

Conclusions

Challenges exist in communicating results to the appropriate personnel on the healthcare team who have the knowledge to act on these data and prescribe targeted therapy against the pathogen(s) identified. In this report, we outline the procedures for telephonic communication and CSP support that were implemented at our health system to distribute RDT data to individuals capable of assessing results, enabling timely optimization of antimicrobial therapy.

Introduction

Bloodstream infections (BSIs) continue to remain a significant cause of morbidity and mortality despite advances in the medical sciences.¹ In 2015, BSIs caused 40 773 deaths in the USA, making them the eleventh leading cause of mortality.¹ Growing rates of antimicrobial resistance put patients at higher risk of morbidity and mortality and lead to increased costs.² BSIs caused by antimicrobial-resistant organisms such as Pseudomonas aeruginosa with difficult-to-treat resistance (DTR), ESBL-producing Enterobacterales, carbapenem-resistant Enterobacterales (CRE), MRSA and VRE are of particular concern.³

Early initiation of appropriate antimicrobial therapy in Gram-negative BSI is important because it has been shown to reduce mortality.^4–6 In patients with septic shock, mortality increases for every hour that adequate antibiotic therapy is not administered.⁷ Early initiation has also been shown to have favourable mortality outcomes in critically ill patients with bacteraemia.^8,9 However, conventional organism identification and susceptibility results typically take 48–72 h to finalize, potentially delaying appropriate antimicrobial therapy.¹⁰ Additionally, this can also delay time to antimicrobial de-escalation, thereby increasing exposure to broad-spectrum agents, potentially leading to resistance, nephrotoxicity, Clostridioides difficile infection and other adverse events.¹¹

Use of rapid diagnostic tests (RDTs) helps identify organisms and their genotypic antimicrobial resistance markers from BSIs within hours of blood culture positivity, potentially allowing for earlier initiation of appropriate antimicrobials.¹² One such test is the BioFire^® FilmArray^® Blood Culture Identification (BCID) Panel (bioMérieux, Marcy l’Étoile), which is a two-stage, multiplex, nested PCR test.

The post-analytical process of interpreting and communicating the results of diagnostic tests is one of the most important factors to their successful implementation in clinical practice. While the improvements in time to clinically important results of a blood culture are a major advancement in the management of BSIs, if these results are not interpreted correctly and acted upon in a timely manner their impact on patient care may be reduced. In fact, the presence of antimicrobial stewardship team (AST) intervention was found to be a key determinant on whether RDTs for BSI had an impact on mortality in a recent meta-analysis.¹³ To go further, the integration of an effective AST coupled with RDTs has demonstrated an improvement of the time to optimal therapy, a decrease in patient length of stay, and total costs for hospitals.¹⁴

To fully demonstrate their value, optimizing the post-analytical phase has become an increasing focus of the implementation process of RDTs in BSI. This includes education, evidence-based guidelines, and use of information technology (IT) tools [e.g. clinical surveillance platforms (CSPs)] for the appropriate personnel on the healthcare team who will interpret RDT results and decide whether to act on these data to modify antibiotic therapy.

Objectives

The aims of this study were to outline the procedural optimization of the post-analytical phase of RDT in our healthcare system and to evaluate if the combination of RDT with a real-time AST support and the use of a CSP would improve the time to appropriate therapy in BSI.

Materials and methods

Study location

This single-centre study was conducted in Cone Health, based in Greensboro, North Carolina, USA. Cone Health is a regional healthcare system with five acute care hospitals, two free-standing emergency departments (EDs), several urgent care centres, and a large network of physician practices. Antimicrobial stewardship has been a focus of Cone Health since 2016 when a pharmacist administrative coordinator was hired. The AST went through two expansions, one in 2019 where two additional pharmacists were hired, and the second in 2021 where another additional pharmacist was hired. The current complement of the AST is one coordinator and three pharmacists. The AST pharmacists perform prospective audit and feedback of antimicrobial use and attend interdisciplinary rounds with the ID consult services where patient care matters are discussed. ID physicians see all patients with Staphylococcus aureus, Enterococcus spp., and Candida spp. BSIs (per hospital policy) and any patient at the request of an attending physician. The microbiology laboratory utilized the BACT/ALERT (bioMérieux) blood culture instrument. Both aerobic and anaerobic blood cultures are routinely utilized. Susceptibility results were obtained using the VITEK 2 (bioMérieux) automated testing system. FDA and CLSI breakpoints were used for interpretation.

Pre-BCID Panel implementation

Prior to implementing the BCID Panel, our team reviewed the available literature to identify key characteristics of successful implementation of RDTs for BSIs. At our healthcare system, positive blood cultures are considered ‘critical’ results. By policy, the laboratory must notify a healthcare team member within 15 min of the positive test. The previous procedure required laboratory staff to contact the patient’s nurse for hospitalized patients. If the patient was discharged, the physician who ordered the test or the patient’s primary care physician was notified. Additionally, our health system required formal infectious diseases (ID) consultation for S. aureus, yeast and enterococcal BSIs. ID consultation was also available for all patients at the request of the attending physician.

BCID Panel implementation

The BCID Panel was first introduced throughout Cone Health in May 2017. BioFire^® FilmArray^® BCID Panel can identify nucleic acid sequences of 24 pathogens and 3 antimicrobial resistance genes including mecA/C for MRSA vanA/B for VRE and bla_KPC for Klebsiella pneumoniae carbapenemase within 1 h. At the time of BCID implementation, the AST consisted of one pharmacist. The VigiLanz^®, an Inovalon solution (Minneapolis, USA) CSP, which receives data from electronic medical records (EMRs), was already in place. Rules were developed to alert the AST pharmacist and other clinical pharmacy staff in real time when a target on BCID was detected.

After collaborating with nursing, laboratory and medical staff leadership, a consensus to change the notification practice (i.e. requirement of laboratory staff to contact the patient’s nurse for hospitalized patients with positive blood cultures) was approved where the laboratory would call a clinical pharmacist with BCID results. To support pharmacists’ decision-making upon receiving a BCID result, a guidance document was created and included recommended actions for escalation, de-escalation, withholding antibiotics, or continuing current treatment (Figure 1). The guidelines provided antibiotic recommendations for first-line agents, alternative therapies in cases of allergy or renal impairment, contact precautions, and if ID consultation would be required per hospital policy. The ID consultant physician’s role was to further guide and refine antibiotic choice, diagnostic testing, duration of antibiotic treatment and hospital follow-up. The pharmacist who received the result from the laboratory would communicate the BCID result along with antibiotic therapy recommendations to the treating physician who was ultimately responsible for the antimicrobial choice. Also, since BCID includes detection of several resistance genes that indicate MRSA, VRE or CRE, clinical pharmacists would place the patient on contact precautions per infection prevention policy.

Figure 1.

Summary of procedural changes required prior to implementation of BCID.

Each day the AST reviewed actions taken for blood cultures that turned positive during the previous 24 h. If the active antibiotic orders did not conform to our guideline recommendations, the AST pharmacist would review the clinical situation and contact the attending physician with suggestions. This was to ensure pharmacists were recommending optimal antibiotics and to track if physicians were accepting the pharmacists’ recommendations.

The expectation was that all clinical pharmacists across all shifts would take BCID calls from microbiology, contact the patient’s physician, provide a treatment recommendation, and document a chart note. The notes included the chief complaint, a working diagnosis, suspected source, physician name, current antibiotics, and recommended changes to antibiotic therapy when applicable. The BCID workflow is shown in Figure 2. A phone list was provided to the microbiology lab, indicating which number to call to reach the clinical pharmacist responsible for responding to the results depending on the time of day and patient location.

Figure 2.

BCID workflow. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Post-BCID Panel implementation

To help reassure physicians that the AST treatment guidelines were appropriate, the AST provided a report to the hospitalist service and the Pharmacy and Therapeutics committee, which included initial empirical regimens selected, the BCID results, pharmacy recommendations on how to optimize therapy, percent acceptance of recommendations, and if the recommendation covered the organism based on final susceptibility results. This report occurred approximately every 6 months for the first 3 years after implementation.

One such evaluation occurred from January 2018 to April 2018. We focused on susceptibility in patients infected with Gram-negative organisms, as genotypic susceptibility for most Gram-positive organisms is present on the BCID Panel. Another reason for the Gram-negative focus was that our institution already had an S. aureus bacteraemia treatment bundle in place, which included mandatory ID consultation, optimizing antibiotics and repeating blood cultures, as well as appropriate imaging and echocardiography. The data collected on this evaluation included the organism, antibiotic(s) on admission, pathogen susceptibility, pharmacist documentation, antibiotic changes based on BCID results and final susceptibility, length of antibiotic therapy, and patient disposition, which included discharge to home, skilled nursing, long-term acute care, hospice, discharge to a group home, transfer to another institution, or death.

Evaluation of BCID2 implementation

The health system transitioned to an expanded, next-generation panel called BCID2 in August 2021. The BioFire^® FilmArray^® BCID2 Panel (bioMérieux) is composed of 14 Gram-negatives, 9 Gram-positives, 7 Candida spp. and 10 resistance genes including CTX-M.

Between August 2021 and March 2022, the times to administration of effective and optimal antibiotics were retrospectively evaluated in patients with the following BCID2 organisms and resistance genes: S. aureus, P. aeruginosa, bla_CTX-M, bla_KPC, bla_OXA-48-like, bla_IMP, bla_VIM, bla_NDM and vanA/B. VigiLanz, our CSP, was instrumental in assisting our AST collect quality assurance data. Specific rules based on BCID2 results were developed in the CSP including real-time notifications to the AST. A CSP report captured critical timepoints including the time blood cultures were drawn, time blood cultures turned positive, time the BCID2 gave a result, antibiotic name(s) and administration time, as well as the patient’s allergy information. The AST defined a list of key performance indicators (KPIs), which included time to effective and optimal antimicrobial therapy, hospital length of stay and patient mortality. The CSP was able to generate the KPIs through a report containing these elements.

An effective antibiotic was defined as an antibiotic with in vitro activity against the organism, and an optimal antibiotic was defined as the preferred agent based on institutional guidelines. A specific analysis was done on the gene encoding CTX-M resistance because this information was not present on the initial BCID Panel but was included on the BCID2 Panel. For CTX-M-positive isolates, we compared time to effective and optimal therapy with a historical control of 34 patients with ceftriaxone-resistant Enterobacterales admitted from January to July 2021 with analysis using Student’s t-test. Genotypic testing was compared with phenotypic results.

Results

Post-BCID Panel implementation

Out of the 101 total organisms evaluated from January 2018 to April 2018, 72% were Escherichia coli, 10% K. pneumoniae, 4% Proteus species, 4% P. aeruginosa, 4% Enterobacter species, 3% Haemophilus influenzae, 3% Klebsiella oxytoca and 1% Serratia marcescens. Initial empirical antibiotics were vancomycin plus piperacillin/tazobactam 32% of the time, followed by ceftriaxone monotherapy at 24%, piperacillin/tazobactam monotherapy at 11%, ceftriaxone plus azithromycin at 7%, and cefepime monotherapy at 6%. Eighty-two patients were evaluated for antibiotic guidance compliance. Upon the pharmacist contacting the physician, antibiotic therapy was promptly changed to coincide with the AST guidance document in 66% (54/82) of patients. Reasons physicians gave for not following AST guidance were variable and included the preference of a different but similar antimicrobial (8/82), the need for anaerobic coverage (3/82), clinical decline (2/82) and concern about resistance (1/82). In several instances, the reason for not accepting recommendations was unknown (14/82).

The final susceptibility of the organism was compared with the AST guidance recommendations and was found to be adequate 98% of the time. The two organisms not covered by initial recommendations were both ESBL-producing Enterobacterales. These data were shared with the AST, the local hospitalist group, the Pharmacy and Therapeutics Committee, and the Medical Executive Committee. Hospital physicians became more comfortable with making antibiotic changes from BCID results over time. Approximately 1 year later, a follow-up evaluation of 350 patients with BSIs with Gram-positive or Gram-negative organisms found that recommendations were rejected only 24 times (6.8%). Reasons for rejection were similar to the previous evaluation. Because of the demonstrated adequacy of antibiotic recommendations and a high percent acceptance of AST recommendations, the AST stopped regular detailed reports in 2020 but continued to follow all patients clinically.

Evaluation of BCID2 implementation

The combined mean time from BCID2 result to administration of effective antibiotics was 1.2 h (range 0–7.9 h) and to optimal therapy was 7.6 h (range 0–113.8 h). In Table 1, the percentage of organisms covered by initial empirical antibiotic therapy was high for MSSA and P. aeruginosa, with 100% coverage. For CTX-M, MRSA and VRE, the percentages were low, with 5%, 40% and 25% of coverage, respectively. The mean time to effective therapy from BCID2 result was less than 3 h for CTX-M-positive organisms (2.8 h), MRSA (2.6 h) and VRE (3.0 h) (Table 1). The time to optimal antibiotics for MSSA was longer (17.0 h ± 24.5 h) as initial empirical therapy was active and there was less urgency to narrow.

Table 1.

Time to effective and optimal antibiotics after implementation of BCID2 (August 2021—March 2022)

Organism and resistance genes identified	Covered by initial empirical antibiotics, % (n/N)	Mean time to effective therapy from BCID2 result to administration of effective therapy for isolates not empirically covered, h (mean ± SD)	Mean time to administration of optimal therapy from BCID2 results, h (mean ± SD)
CTX-M (n = 20)	5 (1/20)	2.8 ± 2.1	2.8 ± 2.0
MSSA (n = 31)	100 (31/31)	0	17.0 ± 24.5
MRSA (n = 25)	40 (10/25)	2.6 ± 1.4	2.0 ± 2.9
VRE (n = 4)	25 (1/4)	3.0 ± 0.8	2.2 ± 1.6
P. aeruginosa (n = 1)	100 (1/1)	0	0
KPC, IMP, VIM, NDM, OXA-48 (n = 0)	N/A	N/A	N/A

After BCID2 implementation, the time to optimal antibiotic therapy for ESBL-producing organisms decreased from 17.7 to 2.8 h (P = 0.0041), as presented in Table 2. One patient with a CTX-M-positive organism and one patient with VRE detected was not confirmed with phenotypic testing.

Table 2.

Mean time to effective therapy from BCID2 result to administration of effective therapy for isolates not empirically covered

	Mean time to effective therapy from BCID2 result to administration of effective therapy for isolates not empirically covered, h (mean ± SD)
Before BCID2—ceftriaxone-resistant isolates	17.7 ± 22.0
After BCID2—CTX-M-positive isolates	2.8 ± 2.1
P value	0.0041

Discussion

Prior to implementing BCID, three important papers were analysed to inform the implementation process.^14–16 A meta-analysis of 31 studies evaluating the impact of RDTs for BSI found mortality was significantly lower with RDTs in the presence of AST support (OR: 0.64, 95% CI: 0.51–0.79).¹⁵ Conversely, when RDTs for BSI were implemented without AST support, there was no significant difference in mortality (OR: 0.72, 95% CI: 0.46–1.12). The second paper was a retrospective, single-centre, cohort study that found an RDT paired with AST support significantly decreased mean hospital length of stay and mean hospital costs compared with an RDT alone (11.9 versus 9.3 days, P = 0.01; $45 709 versus $26 162, P = 0.009).¹⁴ The last paper analysed was a retrospective, single-centre, cohort study where an RDT combined with AST support significantly decreased time to effective therapy compared with conventional organism identification with AST support (17 versus 54 h, P < 0.001).¹⁶ It was clear from this literature review that AST support was needed to drive appropriate and timely antibiotic decisions. In order to deliver higher-quality and more efficient patient care, new IT technologies were developed including data analyses from EMRs and CSP to help ASTs in their daily workflow.¹⁷ Incorporating health informatics allows ASTs to use data from various sources including clinical and microbiological to generate evidence-based recommendations ideally leading to improved patient outcomes.¹⁸ In a meta-analysis of 57 different studies, the authors demonstrated that a CSP was associated with an 18% relative reduction in mortality (OR 0.82), as well as reductions in overall volume of antibiotic use, antibiotic exposure, length of stay and cost of therapy.¹⁹

To replicate the essential aspects of AST support noted in these published studies, we leveraged our existing CSP and current clinical staff to assess and communicate these results. Our health system is fortunate to have a strong clinical pharmacy team, most of whom did at least 1 year of postgraduate pharmacy training and have achieved board certification in pharmacotherapy, critical care or cardiology. Our medical staff rely on our clinical pharmacists for several aspects of acute care, including anticoagulant dosing and monitoring, antimicrobial dosing, renal adjustments, total parenteral nutrition dosing and monitoring, anti-epileptic dosing and monitoring, IV to oral switch for several medications, allergy clarifications and medication reconciliation. Because our clinical pharmacy staff were already well integrated into patient care, the AST felt that clinical pharmacists were ideally positioned to receive these critical results and communicate this information to physicians around the clock. Initially, clinical pharmacy and AST recommendations were taken 66% of the time for Gram-negative organisms, but as physicians became more familiar with the test and its ability to improve antimicrobial stewardship, the acceptance rate increased over time.

The different automatic reports generated by CSP provided our team fast and efficient access to various KPIs, which focused on how quickly patients received effective and optimal antimicrobial therapy. Notably, the addition of the CTX-M resistance marker to BCID2 was highly impactful at our health system, as 95% of patients with an ESBL organism were not adequately empirically covered prior to implementing BCID2. Patients with a history of an ESBL-producing organism are highlighted in our EMR for contact precautions, and we normally recommend empirical carbapenem therapy for this group of patients. However, almost all the patients with an ESBL-producing isolate did not have a prior culture with this organism.

Data showing the reliability and accuracy of the BCID platform coupled with the high likelihood that AST recommendations would cover the detected pathogen were reported to medical staff committees several times in the first few years following implementation. This reassured physicians that the recommendations from pharmacists were trustworthy and a great test to aid in optimizing an antimicrobial regimen through escalation or de-escalation. Prior to BCID implementation in May 2017, de-escalation usually took place when susceptibility results were returned. Having an organism identification and resistance mechanism was a change in practice, and many of our physicians were reluctant to adjust antibiotics based on the BCID result alone. Eventually, the hospitalist group developed confidence in the BCID process to approve clinical pharmacists changing antimicrobial therapy to AST-recommended therapy with the option to opt out of recommendations on a case-by-case basis. The AST was encouraged that recommendations were accepted 93.2% of the time based upon our second evaluation including Gram-positive and Gram-negative organisms.

Ensuring quality care after implementation

The implementation of BCID brought many operational changes to our health system. These changes were positively received by laboratory, pharmacy and physicians, but several unexpected challenges became evident following implementation that needed to be addressed.

Additional workflows needed to be developed for infrequent scenarios such as patients treated and released from the ED whose blood culture subsequently turned positive after discharge from the ED. After collaboration with ED leadership, a designated charge nurse is now contacted by microbiology and communicates culture results to an ED physician to determine the next steps. Actions taken by the ED physician were reviewed by the AST and confirmed for appropriateness. In cases of S. aureus, Enterococcus spp. or yeast, an ID physician was informed via the CSP as well to confirm the appropriateness of care.

There were times when it was difficult for clinical pharmacy staff to reach the attending physician; in these circumstances we would attempt two notifications. If escalation of antimicrobial therapy was not required, no further immediate action would be taken. However, if escalation of antibiotics was needed (e.g. E. coli isolated and patient not on antibiotics), we would contact the on-call ID physician to ensure the patient was started on appropriate antimicrobial therapy.

Another scenario occurred when an organism was seen on Gram stain, but nothing was detected on BCID. For Gram-positive organisms, this finding could represent skin contamination, especially if only one set of blood cultures were positive.²⁰ The need for antibiotics was determined by the attending physician based on the clinical context of the individual patient. If a Gram-negative organism was seen and the patient was not on antibiotics, we recommended that the patient be started on antibiotics as contamination with a Gram-negative organism is less common, and our AST felt that empirical antibiotics should always be considered.²¹ Regardless of the Gram stain and BCID result, the AST would follow the culture until it was finalized and would reach out to an ID physician for expert advice when needed.

Participation of clinical pharmacists in antimicrobial stewardship is recommended by multiple professional organizations.^22–24 However, this participation has traditionally been reserved for clinical pharmacists with additional specialized training (e.g. ID and antimicrobial stewardship specialty training, certificate programmes etc.). There is limited guidance on how or what AST activities front-line pharmacists can perform that lead to improved antimicrobial use and patient outcomes. Through interprofessional development and implementation of AST guidelines, along with real-time antimicrobial decision support, we were able to expand our AST activities by engaging front-line pharmacists in antimicrobial selection of patients with RDT results for BSIs. This intervention resulted in high rates of AST guideline compliance and faster time to effective and optimal therapy in our study population. The use of IT tools such as CSP promptly notified the pharmacists of RDT results with automatic alerts. These CSP alerts allowed AST members to prioritize urgent cases based on predefined criteria such as pathogen type, patient condition and/or antimicrobial resistance patterns.²⁵ Various AST strategies have been attempted, but to date, active real-time AST decision support is the only strategy that has been optimized to provide individualized intervention at the time of rapid diagnostic results.¹⁵ For example, passive notifications of positive blood culture results via EMR alerts did not significantly improve antibiotic prescribing or clinical outcomes among adults admitted to the ICU, highlighting the importance of active interventions to optimize RDT clinical impact.²⁶

While not directly evaluated in the current study, additional barriers to AST activities can be potentially addressed through use of our bundled approach of RDT, AST guidance documents and CSP. For example, automated e-mail and text notifications to AST and ID clinicians can improve the time to de-escalation, ID consultation, contact precautions and initiation of a diagnostic workup sooner than traditional chart and culture review by AST members. Several studies have shown the important role that ID consultation and quality of care process measures can have in the management of antimicrobial-resistant infections.^27–29

Conclusions

The combination of RDT, AST support and IT can greatly aid in optimizing the timing of appropriate antibiotic therapy in patients with BSIs. Clinical pharmacists with appropriate training and support effectively communicated the results and treatment recommendations to the attending physician, allowing for early pathogen-directed antimicrobial therapy. A structure that provides ongoing feedback of microbiological and clinical outcomes is essential to help physicians develop confidence in the test and AST recommendations. CSPs such as VigiLanz^®, an Inovalon solution, are not only helpful in relaying data to the appropriate clinicians but can also generate the high-quality reporting to support such an initiative without manual chart review.

Funding

This paper was published as part of a supplement financially supported by bioMérieux.

Transparency declarations

The authors have no financial conflicts of interest to declare.