Commentary: Can Automated Blood Culture Systems Be Both New and Improved?

ABSTRACT

Automated continuous monitoring blood culture (CMBC) systems are a cornerstone of the clinical microbiology laboratory. Despite the critical role of these systems in diagnosing life-threatening bloodstream infections, their core technologies and performance characteristics have remained largely unchanged since their introduction in the 1990s. This stability and uniformity have enabled the development of quality benchmarks, such as percent positivity and contamination rate; downstream diagnostics, such as direct identification and susceptibility testing of microorganisms in positive cultures; and clinical guidelines based on time to positivity or duration of bacteriemia. In this issue of the Journal of Clinical Microbiology, Chavez et al. (J Clin Microbiol 60:e02261-21. 2021, https://doi.org/10.1128/JCM.02261-21) built on a prior study to examine clinical impacts following the introduction of a new blood culture system which boasts enhanced organism recovery and more rapid time to detection of positive blood cultures. While one might assume that these “improvements” would result in clinical benefits, the authors uncovered some unexpected consequences associated with altering long-accepted performance characteristics. Their central finding was that implementation of the new CMBC system did result in alterations to the management of patients with S. aureus bacteremia; however, this did not have any overall consequences for patient outcomes.

TEXT

Stand-up comedian George Carlin once famously asked the question; “Can something be both new and improved?” The central argument of his rant was that if a product is truly new, then by definition it is not an improvement of an existing product. Likewise, if a product is improved, this implies that the product already existed and therefore is not truly new.

The clinical microbiology laboratory (CML) is currently witnessing an explosion of both truly new technologies and improved existing technologies which are impacting the ways we process specimens, analyze specimens, and report results. These include laboratory automation, digital image analysis, rapid antimicrobial susceptibility testing (AST) methods, highly multiplexed nucleic acid amplification tests (NAATs), and next-generation sequencing, to name a few (1–3). There is little argument that these approaches are “new,” and the data obtained are certainly different that those obtained from existing diagnostic approaches; however, whether or not these technologies are “improvements” is often a subject of debate and may depend on whom you ask.

A prime example is the introduction of matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) as a new technology for routine organism identification, which has undoubtedly been one of the most transformative changes to the CML in the past decade. The shift from biochemical identification methods to MALDI-TOF MS has led to increased accuracy, reduced time to organism identification, and reduced cost per identification; all of which are easily recognizable as improvements from a laboratory point of view. These characteristics have also improved patient care through reductions in time to optimal antibiotic therapy, length of hospital stays, and total cost of care (4, 5). Unfortunately, not all technical advantages of MALDI-TOF MS have translated so seamlessly into clinical and laboratory improvements. The increased accuracy and breadth of organism identification imparted by MALDI-TOF MS have introduced some challenges to the reporting and interpretation of results. For example, isolates previously identified as K. pneumoniae based on biochemical methods may now be reported as K. variicola. While this identification is more accurate, it is likely foreign to providers and introduces discordance with molecular or phenotypic assays which may be conducted on the same specimen (6). This confusion may be further exacerbated by routinely updated MALDI-TOF reference libraries, which add, remove, or edit organism names; coupled with the regulatory requirement for laboratories to report the most current taxonomy (7). Nowhere is this more evident than in the recent revisions to the taxonomic nomenclature of yeasts previously assigned to the genus Candida, many of which now bear unfamiliar names (8) when identified utilizing the most current MALDI-TOF reference libraries.

Management of these unintended consequences (which, at face value, are surely considered improvements) falls to the CML. The clinical benefits of improved accuracy are often first recognized by the clinical microbiology community. Specific examples include the revision of susceptibility testing guidelines for Staphylococcus spp. based on more accurate species-level identification (9) and the recognition of specific clinical presentations associated with certain Corynebacterium spp. (10). However, imparting understanding and application of these changes to patient-facing providers requires a multifaceted approach, including modifications to reporting, continual communication, and education provided by clinical microbiologists, who are best equipped to explain how the changes to laboratory process can or should impact clinical practice.

In this issue of the Journal of Clinical Microbiology, Chavez et al. evaluated a new automated blood culture system, the BacT/Alert VIRTUO (bioMérieux, Durham, NC), to ascertain whether it provides tangible improvements to common laboratory metrics, including positivity rate and time to positivity, compared to their prior system. More importantly, the authors investigated whether these characteristics translated to improvements in clinical care for patients with bloodstream infections (BSI). As in the MALDI-TOF MS example, they encountered some unintended consequences resulting from the intended improvements, which will likely require provider education and modification of existing practice guidelines to ensure that the technical improvements beget improvements to patient care.

Automated continuous monitoring blood culture systems (CMBCs) have been commonplace in CMLs since the introduction of the original BacT/Alert system in 1990. These systems are composed of three basic components: the incubator, the blood culture bottle containing liquid growth medium, and the mechanism to detect the presence of bacterial or fungal growth. Since its initial introduction, the incubator component has remained largely unchanged, while the culture medium and growth detection mechanisms have continued to evolve with the primary goals of providing the highest rate of organism recovery (i.e., sensitivity) and the shortest time to detection, also known as time-to-positivity (TTP). This has been achieved through the optimization of culture medium, the introduction of charcoal or other antibiotic-neutralizing resins to improve recovery of bacteria from patients who may have received antibiotics prior to blood collection, and more sensitive growth detection methods (11).

Over the years, many studies have compared the relative sensitivities and TTP of the three primary CMBC systems and have reported general equivalence, with only minor species-specific differences. Importantly, these performance characteristics have served as the baseline for current guidelines related to the best practices for blood culture collection, including the following metrics (i) optimal volume, (ii) timing, and (iii) utility of follow-up cultures to monitor responses to therapy (12). One such practice guideline published by the Infectious Disease Society of America specifies different clinical management strategies for patients with “persistent” S. aureus bacteriemia, defined as positive blood cultures collected 2 to 4 days after initiation of therapy (13). Recommendations for patients with positive “follow-up” cultures include increased antibiotic dose and duration, and the potential addition of a second antimicrobial agent. Likewise, differential TTP (dTTP) has been proposed as a method for diagnosing catheter-related BSI (CRBSI). This approach is based on a comparison of TTP between blood cultures collected from peripheral venipuncture and those taken from indwelling catheter sites. A dTTP of 2 h is reported to differentiate CRBSI from non-CRBSI with reasonable accuracy compared to other methods (14, 15); these data may be used to inform a clinical decision to remove and replace a central catheter. The CML and developers of in vitro diagnostic tests (IVDs) have similarly relied on the well-established performance features of current CMBC systems (e.g., bacterial density of 10⁷ to 10⁹ CFU/mL in positive cultures) when exploring opportunities to apply new or existing diagnostic methods directly to positive blood culture broths. Such “direct” methods notably include amplified and non-amplified nucleic acid-based pathogen identification panels (e.g., Verigene, FilmArray, ePlex), direct isolation and identification of microorganisms using MALDI-TOF MS (e.g., Sepsityper), and direct phenotypic antibiotic susceptibility testing (e.g., Phenotest, direct disk diffusion).

One can easily imagine the potential downstream impact(s) that may arise if a new (or improved) CMBC system alters one or more of the currently well-established characteristics. For example, the introduction of a more sensitive growth detection technology could reduce TTP by enabling earlier detection of bacterial growth. This, in turn, would result in a lower bacterial density (CFU/mL) at the time of positivity and subsequently reduce the sensitivity of non-amplified nucleic acid detection assays such as Verigene, which has a stated limit of detection of 10⁶ to 10⁷ for many of its targeted organisms (Verigene BC-GP product insert, Rev G). The presence of antibiotics in a blood specimen can have a significant effect on organism recovery and growth kinetics which, in turn, can impact the interpretation and use of TTP. Raad et al. (14) found a 2-h dTTP to be 89% sensitive and 88% specific for the characterization of CRBSI, but the specificity decreased to just 29% among patients who had received antibiotics prior to blood collection. The direct disk diffusion AST method evaluated by the CLSI demonstrated acceptable performance provided that the bacterial density was approximately 10⁸ to 10⁹ CFU/mL at time of positivity; however, authors did not explore the impact of antibiotics on the accuracy of AST results (16). Again, it is easy to envision how more effective neutralization of antibiotics, improvements in culture medium, or advancements in growth-detection technology could alter TTP, growth kinetics, and culture sensitivity, and how these changes to established performance characteristics could require a rethinking of clinical guidelines, post-positivity diagnostics, and quality metrics.

In their evaluation, Chavez et al. focus on detection of S. aureus bacteremia (SAB) because of its relative frequency in causing BSI and the morbidity and mortality associated with secondary complications of infection. When comparing pre- and post-implementation data, the median TTP for initial S. aureus-positive cultures was approximately 4 h shorter when using the “new” VIRTUO system compared to the prior CMBC system. Unquestionably, earlier detection of SAB can be considered a benefit, enabling earlier definitive diagnosis and targeted therapy. A second key finding was that the median duration of “SAB” increased by 2 days (1 day versus 3 days), and the proportion of “prolonged SAB” (defined as lasting >7 days) increased from 4% to 14% of cases, post-implementation of VIRTUO. Closer analysis of the data demonstrated a precipitous decline in positivity following the index culture to just 5% at day 3 in the pre-implementation cohort, while >33% of cultures remained positive at 3 days and 14% remained positive at ≥6 days in the post-implementation cohort, despite initiation of the appropriate antibiotics. The authors were unable to determine whether the increased duration and prolonged recovery time for S. aureus infections was specifically related to the more effective antibiotic neutralization or the more sensitive growth detection conferred by VIRTUO; however, the difference was directly attributable to VIRTUO, given the well-matched demographics of patients in the pre- and post-implementation cohorts. Most notably, certain outcomes, including length of hospital stay, 90-day readmission rate, and 90-day mortality were unchanged. This begs the following question: “Is the increased recovery rate following initial positivity an improvement, neutral, or detrimental to overall patient care?”

To answer this question, the authors compared the utilization frequencies of clinical practices related to SAB management between pre- and post-implementation groups. The use of common imaging studies including transthoracic and transesophageal echocardiograms, often employed to identify patients with endocarditis, was statistically equivalent between groups. In contrast, fluorodeoxyglucose-positron emission tomography/computed tomography (FDG-PET/CT), which is a more expensive and thorough full-body imaging approach, was utilized more frequently in the post-implementation cohort. This increased utilization would be consistent with clinical concerns for an unrecognized nidus of infection, which is potentially responsible for persistently positive cultures. Likewise, antibiotic adjustments broadening therapy to include linezolid, ceftaroline, and multidrug combinations were also significantly more frequent in the post-implementation group; again, this suggests a clinical concern for antibiotic failure or an unrecognized nidus of infection as the cause of persistently positive cultures. Despite these differences in management, the primary clinical and patient outcomes remained unchanged between cohorts.

Based on these data, the authors concluded that the increase in perceived cases of persistent SAB resulting from enhanced recovery by VIRTUO does in fact alter approaches to managing these patients; however, they concluded that these changes did not result in worsening outcomes. While this is a reasonable conclusion, these data should serve as the impetus for a more thorough assessment of impacts beyond readmission rate and all-cause mortality. Potential considerations not examined in this study include the increased adverse effects associated with alternative therapies, higher cost of alternative antimicrobial agents, additional cost and potential risk of unnecessary imaging studies, and additional professional resources, including infectious disease consults for “therapy failure” or “persistent” SAB. Further, as the authors acknowledge, these findings may differ for other institutions depending on which CMBC system they currently use, the institutional guidelines for antibiotic selection, and the evolving literature related to management of patient with SAB.

Most importantly this study serves as a reminder that modifying existing processes or implementing new technologies designed to improve patient care can also result in unforeseen and unintended consequences. Larger studies comparing new or improved CMBC systems are critical to not only determine the analytical and laboratory differences, but also to reevaluate current patient management guidelines and establish other key quality metrics, such as culture positivity and contamination rate, which are tracked longitudinally and may change with the implementation of a new CMBC system. Data are power, but only if they are used. Thorough evaluation of new or improved tests, recognition of potential impacts to patient care, and communication of these changes to our clinical colleagues, is as critical a function of the CML as the generation of reliable results. While George Carlin’s question is a valid one, perhaps a more appropriate quote is provided by American author Norman Cousins: “Wisdom consists of the anticipation of consequences.”

The views expressed in this article do not necessarily reflect the views of the journal or of ASM.