What Do FDA-Cleared Rapid Blood Culture Resistance Tests Detect, and What Do Their FDA Documents Say a Negative Result Does Not Rule out? An Open Device-by-Target Dataset of 24 Molecular and Phenotypic Systems

Abstract

Abstract

Background. Bloodstream infection is a leading cause of death from antimicrobial resistance, and delays in effective therapy cost lives. Two classes of rapid test now report resistance infor-mation directly from positive blood cultures, hours to days before conventional subculture and susceptibility testing: multiplex molecular identification panels, which detect selected resistance genes, and rapid phenotypic antimicrobial susceptibility testing (AST) systems, which measure growth in the presence of antimicrobials. The two classes answer different questions. A detect-ed resistance gene is strong evidence of resistance, but a gene that is not detected does not es-tablish susceptibility, because resistance can arise through mechanisms a panel does not target. Phenotypic systems report susceptibility directly, but only for the antimicrobial and organism combinations their authorization covers. These boundaries are described in each device's FDA documentation, yet that documentation is spread across dozens of decision summaries written over nearly two decades in different formats. A laboratory choosing a rapid test, writing result comments, or verifying a new system under federal laboratory rules therefore has no single public source showing what each FDA-authorized test detects and what its FDA documenta-tion says a negative resistance result does and does not rule out. Methods. We conducted a cross-sectional analysis of public U.S. regulatory records. Device records were identified through the openFDA 510(k) interface (data current to 26 October 2026) in nine product codes covering positive blood culture identification panels, rapid AST systems and tests for S. aureus/methicillin-resistant S. aureus (MRSA), supplemented by a device-name search. Records were included if the test is performed directly on positive blood culture and reports at least one resistance gene, a resistance phenotype or antimicrobial susceptibility. Submissions for the same assay were grouped into device families, and one FDA document per family, normally the most recent decision summary restating the full indications for use, was abstracted. For each family we recorded organism and resistance targets as written, the organ-isms with which each resistance result is reported or attributed, every antimicrobial and organ-ism combination in the AST indications for use, specimen requirements, and labeled limitations transcribed verbatim and classified into 16 predefined types. Every value is traced to its sub-mission number and page. The primary outcome was whether each molecular or MRSA-screen document states that a negative resistance-marker result does not indicate susceptibility. As a pre-specified secondary comparison, documents dated before and from 2020 were compared with Fisher's exact test. Before release, all openFDA queries were re-run, every submission field was checked against live records, extracted data were reviewed against the source docu-ments, and all derived files were reproduced from a fresh copy. Results. Of 72 records screened, 40 submissions (38 510(k) clearances and 2 De Novo authoriza-tions, dated 2007-12-20 to 2026-09-24) from 24 device families were included: 12 multiplex mo-lecular panels, 5 molecular S. aureus/MRSA tests, 1 phenotypic MRSA screen and 6 rapid phe-notypic AST systems. Rapid phenotypic AST is recent: 10 of its 12 submissions were decided in 2024 to 2026. 16 documents tie use to a Gram-stain result. The documents named 217 organism targets and 90 resistance targets (83 genes and 7 qualitative phenotypes) in 28 marker families. Breadth ranged from 1 resistance target on single-organism S. aureus tests to 18 on the broadest Gram-negative panel. Mec genes were most widely represented (11 devices), followed by KPC and van genes (7 each) and CTX-M and NDM (6 each). Only 4 Gram-negative panels named all five carbapenemase families most often reported in the United States (KPC, NDM, VIM, IMP and OXA-48-like). 16 marker families appeared on a single device, 13 of them on two panels authorized in 2026. Attribution of resistance results varied: three documents report a gene only with an associated organism, four state that genes are not attributed in mixed growth, 9 warn that a detected gene may not belong to the causative organism, and for 6 panels the indications do not say which organisms a resistance result is reported with. Of 18 molecular and MRSA-screen documents, 9 stated that a negative resistance-marker result does not indicate suscepti-bility (one with hedged wording) and 9 did not (one only by indirect reference to the package insert). All 6 documents dated 2020 or later carried the statement, compared with 3 of 12 dated earlier (Fisher's exact p = 0.009). The documents without it, dated 2007-12-20 to 2019-06-28, in-clude all four tests limited to S. aureus/MRSA. 4 documents also warned that a negative result may reflect a pathogen the panel does not target. The 6 phenotypic AST systems listed 841 an-timicrobial and organism indications covering 32 antimicrobials. 7 agents were indicated on all six systems (amikacin, cefepime, ceftazidime, ertapenem, gentamicin, meropenem, piperacillin-tazobactam), whereas ceftazidime-avibactam was indicated on 5, meropenem-vaborbactam on 3 and ceftolozane-tazobactam on 2. Only two systems covered Gram-positive organisms. Of 292 normalized organism and antimicrobial pairs, 73 (25.0%) were indicated on only one system and 19 (6.5%) on all six. 5 systems required organism identification from a separate method, 5 stated that results are not valid for polymicrobial samples, 4 listed combinations requiring an al-ternate method before reporting, and 4 listed combinations whose ability to detect resistance is unknown because too few resistant isolates were studied. All 24 documents stated that subcul-ture of the positive blood culture is still required. Conclusions. FDA-authorized rapid blood culture resistance tests differ substantially in what they detect and in what their FDA documents say about a negative result. Documents dated be-fore 2020 often omit the warning that a negative resistance-marker result does not indicate sus-ceptibility, so laboratories using older tests should check their labeling and consider adding that warning to their own result comments. Resistance gene attribution is limited in mixed cultures, and rapid phenotypic AST coverage remains fragmented across systems, with a quarter of or-ganism and antimicrobial pairs available on only one platform. These findings describe FDA documents rather than current package inserts or test performance, and the era comparison is exploratory. The dataset, code, comparison tables and provenance log are openly released to support test selection, verification and result reporting, and will be refreshed as new devices are authorized.