FDA 510(k) Search - By Innolitics (SaMD and AI Experts)

HardyCHROM ESBL is a selective and differential chromogenic medium which is intended for the qualitative and presumptive detection from stool specimens of: 1) Enterobacteriaceae that are potentially non-susceptible to ceftazidime and cefpodoxime; and 2) Extended-spectrum beta-lactamase (ESBL)-producing Escherichia coli, Klebsiella pneumoniae and Klebsiella oxytoca.

The test is performed on stool specimens from patients at risk of harboring Enterobacteriaceae that are non-susceptible to 3rd generation cephalosporins or ESBL-producing E. coli, K. pneumoniae and K. oxytoca, and is intended as an aid in the detection, identification of colonization and control of these bacteria in a healthcare setting. HardyCHROM ESBL is not intended to diagnose infection or to guide or monitor treatment for infections. Results can be interpreted after incubation for 18-24 hours. Subculture to non-selective medium is required for confirming identification, susceptibility testing and epidemiological typing.

A lack of growth or the absence of pink, blue or yellow/gold colonies on HardyCHROM ESBL does not preclude the presence of Enterobacteriaceae that are non-susceptible to 3rd generation cephalosporins or ESBL producing organisms.

Device Description

HardyCHROM™ ESBL is a selective and differential chromogenic medium containing a broad-spectrum beta-lactam intended for detection and isolation of Enterobacteriaceae non-susceptible to 3rd generation cephalosporins. HardyCHROM™ ESBL can also be used as a screening medium for K. pneumoniae, K. oxytoca, and E. coli that produce an extended-spectrum beta-lactamase (ESBL).

The selective components in HardyCHROM™ ESBL are designed to inhibit the growth of yeast, Grampositive bacteria, and Gram-negative bacteria sensitive to broad spectrum beta-lactams (3rd generation cephalosporins). Chromogenic substrates in the medium allow for differentiation of Enterobacteriaceae non-susceptible to 3rd generation cephalosporins or that are ESBL-producing, as bacteria that can grow and utilize the chromogens produce a colored colony. ESBL-producing Klebsiella spp. produce large, dark blue colonies. ESBL-producing Escherichia coli produce colonies that are rose to magenta in color, with darker pink centers. Other Enterobacteriaceae not susceptible to 3rd generation cephalosporins will produce colonies of varying size that are pink, blue, with pink halos, and yellow/gold.

AI/ML Overview

The provided document is a 510(k) Premarket Notification for the HardyCHROM ESBL device. It details the device's indications for use, its comparison to a predicate device, and performance data from various studies.

Here's an analysis of the acceptance criteria and the study proving the device meets them, based on the provided text:

Device: HardyCHROM™ ESBL
Intended Use: Qualitative and presumptive detection from stool specimens of:

Enterobacteriaceae potentially non-susceptible to ceftazidime and cefpodoxime.
Extended-spectrum beta-lactamase (ESBL)-producing Escherichia coli, Klebsiella pneumoniae, and Klebsiella oxytoca.

1. Table of Acceptance Criteria and Reported Device Performance

The document doesn't explicitly state "acceptance criteria" as a separate, pre-defined set of thresholds. Instead, it presents performance metrics from various studies (clinical, contrived, analytical reactivity, analytical specificity, microbial interference, specimen stability, and reproducibility). The results are presented as the device's performance, implying these values met the internal requirements for substantial equivalence.

Here's a table summarizing key performance metrics observed in the studies:

Performance Metric Category	Specific Metric	Reported Device Performance	Acceptance Criteria (Implied/Discussed)
Clinical Performance (Prospective)	Detection of 3rd gen. cephalosporin non-susceptible Enterobacteriaceae	Overall Sensitivity: 91.0% (95% CI: 87.8% - 93.4%)	While not explicitly stated as acceptance criteria, these sensitivity and specificity values are presented as evidence of effective performance compared to traditional culture. For a screening device, high sensitivity is often prioritized to minimize false negatives, while reasonable specificity is needed to avoid excessive false positives. The CIs provide an indication of the precision of these estimates.
	Overall Specificity: 95.2% (95% CI: 94.1% - 96.1%)
	Detection of ESBL-producing K. oxytoca, K. pneumoniae, E. coli	Overall Sensitivity: 97.1% (95% CI: 93.3% - 98.7%)	Similar to above, high sensitivity is crucial for detecting ESBL producers, given their clinical significance. The very high sensitivity (97.1%) and the fact that one site achieved 100% sensitivity for this indication suggest strong performance. Specificity is also important to prevent unnecessary further testing.
	Overall Specificity: 87.1% (95% CI: 85.6% - 88.5%)
Contrived Specimen Performance	PPA (Presumptive Detection of Ref. NS Enterobacteriaceae)	97.9% (47/48) (95% CI: 89.1% - 99.6%)	The high PPA and NPA (where applicable) for contrived specimens, particularly at LoD (Limit of Detection), indicate the consistent performance of the device under controlled conditions and its ability to detect target organisms even at low concentrations. The recovery rate (LoD) at 10^3 CFU/mL or 10 CFU/plate suggests the device is sensitive enough for clinical utility.
	NPA (Presumptive Detection of Ref. NS Enterobacteriaceae)	60.0% (3/5) (95% CI: 23.1% - 88.2%)	(Note: The NPA for presumptive detection of non-susceptible Enterobacteriaceae on contrived specimens is low, likely due to the small sample size (n=5 negative contrived samples) and high variability inherent in such small numbers. This specific metric's low value might be considered acceptable due to the limited sample, the overall positive performance for ESBL, and the requirement for subculture for confirmation.)
	PPA (Contrived ESBL-"EC" Pink)	100% (19/19) (95% CI: 83.1% - 100%)
	NPA (Contrived ESBL-"EC" Pink)	91.2% (31/34) (95% CI: 77.0% - 97.0%)
	PPA (Contrived ESBL-"KP/KO" Blue)	100% (26/26) (95% CI: 87.1% - 100%)
	NPA (Contrived ESBL-"KP/KO" Blue)	92.6% (25/27) (95% CI: 76.6% - 97.9%)
Recovery Rate (LoD)	3rd gen. cephalosporin non-susceptible Enterobacteriaceae	No discernable difference in recovery at 10^3 CFU/mL (stool)	The ability to recover target organisms at low concentrations (LoD) is crucial for a screening device. This indicates the device is sufficiently sensitive.
	ESBL-producing E. coli, K. oxytoca, K. pneumoniae	No discernable difference in recovery at 10 CFU/plate (stool)
Analytical Reactivity	Recovery of 54 characterized ESBL-producing strains	100% (54/54) after 24 hours of incubation	100% recovery indicates the device's ability to detect a broad range of ESBL-producing strains with various genotypes. This is a critical analytical performance metric.
	Recovery of 21 3rd gen. cephalosporin non-susceptible strains	100% (21/21) after 24 hours of incubation (71.4% with expected color)	High recovery (100%) for non-susceptible strains is important. The note about 71.4% with expected color indicates that some non-ESBL non-susceptible organisms might not show the typical chromogenic reaction, but are still detected. This aligns with the "presumptive" nature.
Analytical Specificity	Non-growth/non-target color for non-pathogens/susceptible organisms	Majority of non-Enterobacteriaceae did not grow or exhibit target colors.	The device should be specific to the target organisms to minimize false positives and reduce the need for unnecessary follow-up. High specificity (inhibiting non-target growth or showing non-target color) is key for a selective medium.
	3rd gen. cephalosporin non-susceptible Enterobacteriaceae (non-ESBL)	100% (45/45) either no growth or colors other than pink/blue/yellow-gold.	This indicates good specificity in distinguishing ESBL-producing organisms (which are expected to show target colors) from other non-susceptible Enterobacteriaceae, which may not.
Reproducibility	PPA for detection of 3rd gen. cephalosporin non-susceptible Enterobacteriaceae	Overall PPA: 98.2% (95% CI: 95.7% - 99.2%)	High reproducibility across different sites and operators is essential to ensure consistent performance in various clinical settings. PPA and NPA values close to 100% demonstrate this consistency.
	NPA for detection of 3rd gen. cephalosporin non-susceptible Enterobacteriaceae	Overall NPA: 99.6% (95% CI: 98.0% - 99.9%)
	PPA for detection of ESBL-producing K. oxytoca, K. pneumoniae, E. coli	Overall PPA: 98.1% (95% CI: 95.3% - 99.3%)
	NPA for detection of ESBL-producing K. oxytoca, K. pneumoniae, E. coli	Overall NPA: 83.3% (95% CI: 79.0% - 86.9%)	The lower NPA here is noted. The document explains that false positives included E. cloacae (a non-ESBL producing organism that can still show blue colonies on HardyCHROM ESBL if it’s carbapenem resistant or has other beta-lactamase mechanisms like AmpC). If these are omitted, NPA improves to 99.6%. This indicates that while the device is highly sensitive, interpretation requires confirmation for full specificity. The "presumptive detection" in the Indications for Use allows for this, as subculture is required for confirmation.

2. Sample Size Used for the Test Set and Data Provenance

Clinical Study (Prospective):

Sample Size: 1,559 valid clinical stool specimens (out of 1,687 initially tested).
Data Provenance: Fresh stool specimens collected prospectively at three geographically diverse hospitals. The specific countries are not mentioned, but the context implies US given the FDA submission.

Contrived Specimen Study:

Sample Size: 50 contrived specimens.
Data Provenance: Patient specimens inoculated with known resistant organisms (ESBL producing or non-susceptible to third-generation cephalosporins) at the Limit of Detection (LoD).

Reproducibility Study:

Sample Size:
- For 3rd gen. cephalosporin non-susceptible Enterobacteriaceae: 272 positive samples (N) and 280 negative samples (N) overall across 5 days of testing and 3 sites.
- For ESBL-producing organisms: 216 positive samples (N) and 336 negative samples (N) overall across 5 days of testing and 3 sites.
Data Provenance: Blinded panels of bacterial suspensions in transport medium. Tested at three different sites.

Analytical Reactivity & Specificity Studies:

Analytical Reactivity (ESBL): 54 characterized ESBL-producing strains of E. coli, K. pneumoniae, and K. oxytoca.
Analytical Reactivity (Non-ESBL non-susceptible): 21 strains representative of various Enterobacteriaceae that were non-susceptible to at least one 3rd generation cephalosporin.
Analytical Specificity: 99 strains (52 Enterobacteriaceae and 47 non-Enterobacteriaceae, including ESBL non-producing and 3rd gen. cephalosporin susceptible species).
Data Provenance: Internal testing using characterized bacterial strains.

3. Number of Experts Used to Establish Ground Truth for the Test Set and Qualifications

For the Clinical Study:

Ground Truth Establishment: "Routine culture, defined as the selective enrichment of microorganisms in Tryptic Soy Broth (TSB) containing either 1ug/mL ceftazidime or 1ug/mL cefotaxime, followed by subculture to MacConkey Agar. Organisms that grew on MacConkey Agar were identified using an FDA-cleared ID system. Confirmation of ESBL production and 3rd generation cephalosporin non-susceptibility was performed using traditional Kirby-Bauer AST following the device manufacturer's instructions. ID of organisms that grew on HardyCHROM™ ESBL was confirmed using an FDA-cleared ID system."
Number/Qualifications of Experts: The document does not specify the number or qualifications of experts involved in performing the "routine culture" or interpreting the Kirby-Bauer AST and FDA-cleared ID system results. This would typically be performed by trained laboratory personnel (e.g., medical technologists, clinical microbiologists).

For the Reproducibility Study:

Ground Truth Establishment: Not applicable as it's a test of reproducibility using known positive and negative controls.
Number/Qualifications of Experts: Readings were performed by 2 independent operators at each of the three sites (total 6 operators). Their specific qualifications beyond "operator" are not detailed.

4. Adjudication Method for the Test Set

For the Clinical Study:

The ground truth seems to be established by a defined laboratory methodology (selective enrichment, subculture, FDA-cleared ID, Kirby-Bauer AST). There is no explicit mention of expert adjudication of the ground truth in the clinical study beyond the standard laboratory procedures. It's a method-based ground truth comparison.

For the Reproducibility Study:

Readings were done by "2 independent operators." There is no explicit mention of an adjudication process if their readings differed. However, for a reproducibility study with known controls, discrepancies might lead to re-testing or investigation rather than adjudication of the "truth."

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was done

No, a MRMC comparative effectiveness study was not done in the context of comparing human readers with and without AI assistance, as this device is a culture medium, not an AI or imaging diagnostic tool. The performance is compared against a "traditional culture" method.
The reproducibility study involved multiple readers (operators) across multiple sites, but it was for assessing inter-operator and inter-site variability of the device itself, not a comparative effectiveness study involving AI assistance for human readers.

6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) was done

Yes, in essence, standalone performance was evaluated. The device (HardyCHROM™ ESBL) is a culture medium, and its "standalone" performance is measured by its ability to recover and differentiate target organisms based on colony growth and color from the specimen. Human observation is then used to interpret these results.
The sensitivity and specificity data presented for the clinical and analytical studies reflect the performance of the medium itself to appropriately grow and present the target organisms, which is then visually interpreted. There is no "algorithm" separate from the biological reaction on the medium.

7. The Type of Ground Truth Used

Clinical Study: The ground truth for the clinical study was established by conventional microbiological methods, including selective enrichment, subculture, identification using FDA-cleared ID systems, and confirmation of non-susceptibility/ESBL production using traditional Kirby-Bauer Antimicrobial Susceptibility Testing (AST). This is a method-based ground truth derived from established laboratory practices.
Contrived Specimen Study: The ground truth was based on known resistant organisms (ESBL-producing or non-susceptible) used to inoculate patient specimens. This is a "spiked" study design where the ground truth is precisely controlled.
Analytical Reactivity/Specificity: The ground truth for these studies was based on characterized bacterial strains with known ESBL production, non-susceptibility, or other relevant characteristics (e.g., non-ESBL, susceptible, non-Enterobacteriaceae).

8. The Sample Size for the Training Set

The document does not mention a "training set" in the context of machine learning or AI models. This device is a culture medium relying on biological and chemical reactions, not an AI algorithm that requires training data.
The studies described are primarily for performance validation/verification of the manufactured product (culture medium).

9. How the Ground Truth for the Training Set was Established

As there is no AI algorithm training set, this question is not applicable to this device.

Summary

Regulation Number and Section

§ 866.1700 Culture medium for antimicrobial susceptibility tests.

(a)
Identification. A culture medium for antimicrobial susceptibility tests is a device intended for medical purposes that consists of any medium capable of supporting the growth of many of the bacterial pathogens that are subject to antimicrobial susceptibility tests. The medium should be free of components known to be antagonistic to the common agents for which susceptibility tests are performed in the treatment of disease.(b)
Classification. Class II (performance standards).