(57 days)
COBAS INTEGRA Ammonia (NH3): contains an in vitro diagnostic reagent system intended for use on COBAS INTEGRA for the quantitative determination of the ammonia concentration in plasma (test NH3, 0-045).
COBAS INTEGRA aAmylase EPS (AMYLL): contains an in vitro diagnostic reagent system intended for use on COBAS INTEGRA for the quantitative determination of the catalytic activity of amylase in serum, plasma (test AMY-L, 0998) and urine (test AMY-UL 0-999).
COBAS INTEGRA Cholesterol (CHOLL): contains an in vitro diagnostic reagent system intended for use on COBAS INTEGRA for the quantitative determination of total cholesterol (test CHOLL, 0-001) and HDL cholesterol concentration in serum and plasma in clinical laboratories.
COBAS INTEGRA HDL Cholesterol Application (HDLL): contains an in vitro diagnostic reagent system intended for use on COBAS INTEGRA for the quantitative determination of total cholesterol and HDL - cholesterol (test HDLL, 0-002) concentration in serum and plasma in clinical laboratories.
COBAS INTEGRA Creatinine Enzymatic (CREAE): contains an in vitro diagnostic reagent system intended for use on COBAS INTEGRA for the quantitative determination of the creatinine concentration in serum (test CREAE, 0-014), and urine (test CREEU, 0-114).
COBAS INTEGRA Digitoxin (DIGIT): contains an in vitro diagnostic reagent system intended for use on COBAS INTEGRA for the quantitative determination of digitoxin in serum or heparinized plasma (test DIGIT 0-259).
COBAS INTEGRA Gamma Glutamyltransferase (GGTL): contains an in vitro diagnostic reagent system intended for use on COBAS INTEGRA for the quantitative determination of the catalytic activity of GGT, (EC 2.3.2.2; y-glutamyl peptide: amino acid y-glutamyltransferase) in serum and plasma (test GGTL, 0-599).
COBAS INTEGRA Glucose HK Liquid (GLUCL): contains an in vitro diagnostic reagent system intended for use on COBAS INTEGRA for the quantitative determination of the glucose concentration in serum, plasma (test GLUL, 0-991), urine (test GLULU, 0-992), and cerebrospinal fluid (test GLULC, 0-993).
COBAS INTEGRA Lipase (LIPL): contains an in vitro diagnostic reagent system intended for use on COBAS INTEGRA for the quantitative determination of the catalytic activity of lipase in serum and plasma (test LIPL, 0-200).
COBAS INTEGRA Lysergic acid diethylamide (LSD) contains an in vitro diagnostic reagent system intended for use on COBAS INTEGRA for the qualitative determination of lysergic acid diethylamide (LSD) in urine (test LSD, 0-001)
COBAS INTEGRA Urea/BUN (UREAL): contains an in vitro diagnostic reagent system intended for use on COBAS INTEGRA for the quantitative determination of the urea/BUN (blood urea nitrogen), in serum, plasma (test UREL, 0-003) and urine (test URELU, 0-004).
Roche TDM OnLine Digitoxin Calibrators: are intended for use with the Roche reagents for Digitoxin and the COBAS Chemistry systems for the quantitative determination of digitoxin in serum and plasma.
Roche TDM OnLine Digitoxin Controls: are quality control samples intended for use on COBAS chemistry systems with Roche reagents and calibrators for the quantitative determination of digitoxin assays.
The COBAS INTEGRA test applications contained in this submission are intended for use with the COBAS INTEGRA Analyzer. The COBAS INTEGRA Analyzer and COBAS INTEGRA Reagent cassettes together provide an integrated system for in vitro diagnostic testing. The COBAS INTEGRA Analyzer utilizes three measuring principles, i.e., absorbance, fluorescence polarization and ion-selective electrodes. The analyzer has a throughput of up to 600 tests per hour with STAT samples prioritized and tested immediately. Random sample access, robotics and a user interface optimize time management and streamline workflow. The COBAS INTEGRA can store up to 68 COBAS INTEGRA Reagent Cassettes on board, 24 hours a day at 2-8℃. The COBAS INTEGRA Reagent Cassettes are compact and preparation-free with the added convenience of long term on-board stability. Barcode readers are used to identify newly loaded reagent cassettes, samples for patient identification, and rack inserts and to read calibration and control data from the cassette label. COBAS INTEGRA tests include chemistry, drugs of abuse, immunology, ion selective electrodes, therapeutic drug monitoring, and hematology reagents. Through this submission, it is the intention of Roche Diagnostic Systems to gain clearance for an additional 4 COBAS INTEGRA Reagent Cassettes and 2 ancillary reagents as well as modifications to 7 previously cleared COBAS INTEGRA Reagent Cassettes. These reagents have been modified from granulate to liquid form.
The provided 510(k) summary (K972250) describes the acceptance criteria and study results for several Roche COBAS INTEGRA Reagent Cassettes and ancillary reagents. The studies are primarily focused on demonstrating substantial equivalence to predicate devices, rather than establishing de novo performance criteria against a fixed clinical standard. Consequently, the "acceptance criteria" are implied by the results of the comparative studies to be within acceptable analytical performance limits for equivalent devices.
Here's a breakdown of the requested information for each reagent, based on the provided text:
Roche COBAS® INTEGRA Reagent Cassettes & Ancillary Reagents (K972250)
The acceptance criteria are generally implied by the strong correlation and similar performance characteristics (assay range, precision, sensitivity, accuracy/linearity) when compared to the legally marketed predicate devices. The study's goal was to demonstrate substantial equivalence, meaning the new device performs comparably to the predicate.
1. Table of Acceptance Criteria and Reported Device Performance
The acceptance criteria for these in vitro diagnostic devices are demonstrated through a comparison of their performance characteristics (Assay Range, Precision, Sensitivity, Accuracy/Correlation Coefficient, and Linear Regression) against legally marketed predicate devices. The "reported device performance" is the performance of the COBAS INTEGRA (Liquid) reagents. The "acceptance criteria" are implied to be within comparable ranges to the predicate devices, indicating substantial equivalence.
Note: For each test, the predicate device's performance is presented alongside the new device's performance, and the linear regression typically shows correlation against the predicate. This comparative approach is the core of the acceptance criteria.
Ammonia (NH3)
| Performance Characteristic | Acceptance Criteria (Predicate) | Reported Device Performance (COBAS INTEGRA Ammonia Liquid) |
|---|---|---|
| Assay Range | 0 - 700 U/L (0-2800 U/L with postdilution) | 0 - 700 µmol/L (0-1190 ug/dL) |
| Precision (Level 1) | Mean: ૯દિવ; %CV (within run): 5.7; %CV (total): 8.8 | Mean: 48.8 µmol/L; %CV (w/r): 3.1; %CV (total): 5.2 |
| Precision (Level 2) | Mean: 211; %CV (within run): 1.9; %CV (total): 5.9 | Mean: 226 µmol/L; %CV (w/r): 2.0; %CV (total): 2.5 |
| Sensitivity | 0.0009 AA per µmol/L | 0.76 AA per µmol/L |
| Accuracy (n=164) | Corr. Coefficient (r): 0.992 | Corr. Coefficient (r): 0.997 |
| Lin. Regression | 1.02x + 3.2 µmol/L | 1.03x - 2.8 µmol/L vs. Roche Reagent for Ammonia |
Creatinine (CREAE) - Serum and Plasma
| Performance Characteristic | Acceptance Criteria (Predicate: COBAS INTEGRA Creatinine (Kinetic, Jaffé)) | Reported Device Performance (COBAS INTEGRA Creatinine (Enzymatic, PAP)) |
|---|---|---|
| Assay Range | 0 - 1300 µmol/L (0-13000 µmol/L with post dilution) | 0 - 2000 µmol/L (0-20000 µmol/L with post dilution) |
| Precision (Level 1) | Mean: 85.5 µmol/L; %CV (w/r): 1.5; %CV (total): 1.9 | Mean: 99.4 µmol/L; %CV (w/r): 1.6; %CV (total): 2.2 |
| Precision (Level 2) | Mean: 624 µmol/L; %CV (w/r): 1.1; %CV (total): 1.5 | Mean: 535 µmol/L; %CV (w/r): 0.88; %CV (total): 1.5 |
| Sensitivity | 8.0 X 10^-5 ΔA/min per µmol/L | 2.2 X 10^-4 ΔA per µmol/L |
| Accuracy (n=238) | Corr. Coefficient (r): 0.999 | Corr. Coefficient (r): 0.999 |
| Lin. Regression | 0.87x - 2 µmol/L | 1.08x - 30.6 µmol/L vs. COBAS INTEGRA Creatinine (Jaffé) |
Creatinine (CREAE) - Urine
| Performance Characteristic | Acceptance Criteria (Predicate: COBAS INTEGRA Creatinine (Kinetic, Jaffé)) | Reported Device Performance (COBAS INTEGRA Creatinine (Enzymatic, PAP)) |
|---|---|---|
| Assay Range | 0 - 32.5 mmol/L (0-130 mmol/L with post dilution) | 0 - 40 mmol/L (0-200 mmol/L with post dilution) |
| Precision (Level 1) | Mean: 5.3 mmol/L; %CV (w/r): 1.5 | Mean: 4.1 mmol/L; %CV (w/r): 0.88; %CV (total): 1.1 |
| Precision (Level 2) | Mean: 19 mmol/L; %CV (w/r): 1.0 | Mean: 14.0 mmol/L; %CV (w/r): 0.87; %CV (total): 0.93 |
| Sensitivity | Not specified in labeling | 5.7 X 10^-3 ΔA per mmol/L |
| Accuracy (n=116) | Not specified in labeling | Corr. Coefficient (r): 0.999 |
| Lin. Regression | Not specified in labeling | 0.99x - 0.28 mmol/L vs. COBAS INTEGRA Creatinine (Jaffé) |
Digitoxin (DIGIT)
| Performance Characteristic | Acceptance Criteria (Predicate: Abbott TDx/TDxFLx Digitoxin) | Reported Device Performance (COBAS INTEGRA Digitoxin) |
|---|---|---|
| Assay Range | 2.0 - 80 ng/mL | 2.0 - 65 ng/mL |
| Precision (Level 1) | Mean: 7.5 ng/mL; %CV (w/r): 7.05; %CV (total): 10.61 | Mean: 10.4 ng/mL; %CV (w/r): 6.0; %CV (total): 7.4 |
| Precision (Level 2) | Mean: 15.0 ng/mL; %CV (w/r): 4.87; %CV (total): 7.19 | Mean: 19.5 ng/mL; %CV (w/r): 3.9; %CV (total): 4.5 |
| Precision (Level 3) | Mean: 35.0 ng/mL; %CV (w/r): 4.72; %CV (total): 8.46 | Mean: 37.1 ng/mL; %CV (w/r): 3.6; %CV (total): 3.7 |
| Sensitivity | 2.0 ng/mL | 2.0 ng/mL |
| Accuracy (n=232) | Corr. Coefficient (r): 0.967 | Corr. Coefficient (r): 0.973 |
| Lin. Regression | 1.060 + 0.729 ng/mL | 0.945x + 1.19 ng/mL vs. Abbott TDx/TDxFLx Digitoxin |
Lysergic acid diethylamide (LSD)
| Performance Characteristic | Acceptance Criteria (Predicate: Roche Abuscreen RIA for LSD) | Reported Device Performance (COBAS INTEGRA LSD) |
|---|---|---|
| Assay Range | 0 - 1 ng/mL | 0 - 1 ng/mL |
| Precision (Level 1) | Mean: 0.0; %CV (w/r): 0.6 | Mean (O.D.): 0.978; %CV (w/r): N/A |
| Precision (Level 2) | Mean: 0.25; %CV (w/r): 1.3 | Mean (O.D.): 0.913; %CV (w/r): N/A |
| Precision (Level 3) | Mean: 0.5; %CV (w/r): 1.6 | Mean (O.D.): 0.870; %CV (w/r): N/A |
| Sensitivity | 0.25 ng/mL of LSD at > 99% confidence | 0.10 ng/mL of LSD at > 95% confidence |
| Accuracy | Positive Samples (GC/MS): 21/0; Positive Samples (RIA): 21/0 | Positive Samples (GC/MS): 39/0; Positive Samples (RIA): 39/0 |
Note: For LSD, precision is presented in Optical Density (O.D.) for the new device vs. ng/mL for the predicate, making direct comparison of mean values challenging. However, the %CV for within-run are similar.
α-Amylase (AMYLL) - Serum and Plasma
| Performance Characteristic | Acceptance Criteria (Predicate: COBAS INTEGRA α-Amylase (Granulate)) | Reported Device Performance (COBAS INTEGRA α-Amylase EPS (Liquid)) |
|---|---|---|
| Assay Range | 0 - 2000 U/L (0-20000 U/L with post dilution) | 0 - 2000 U/L (0-10000 U/L with post dilution) |
| Precision (Level 1) | Mean: 143 U/L; %CV (w/r): 1.6; %CV (total): 1.6 | Mean: 76 U/L; %CV (w/r): 1.6; %CV (total): 2.3 |
| Precision (Level 2) | Mean: 277 U/L; %CV (w/r): 1.1; %CV (total): 2.0 | Mean: 498 U/L; %CV (w/r): 1.3; %CV (total): 2.6 |
| Sensitivity | 1.5 X 10^-4 ΔA/min per U/L | 1.9 X 10^-4 ΔA/min per U/L |
| Accuracy (n=114) | Corr. Coefficient (r): 0.992 | Corr. Coefficient (r): 0.996 |
| Lin. Regression | 0.98x - 19 U/L | 0.43x + 4 U/L vs. COBAS INTEGRA α-Amylase (granulate) |
α-Amylase (AMYLL) - Urine
| Performance Characteristic | Acceptance Criteria (Predicate: COBAS INTEGRA α-Amylase (Granulate)) | Reported Device Performance (COBAS INTEGRA α-Amylase EPS (Liquid)) |
|---|---|---|
| Assay Range | 0 - 2000 U/L (0-20000 U/L with post dilution) | 0 - 2000 U/L (0-10000 U/L with post dilution) |
| Precision (Level 1) | Mean: 22 U/L; %CV (w/r): 2.5 | Mean: 183 U/L; %CV (w/r): 1.3; %CV (total): 1.7 |
| Precision (Level 2) | Mean: 302 U/L; %CV (w/r): 0.56 | Mean: 603 U/L; %CV (w/r): N/A; %CV (total): 1.6 |
| Sensitivity | Not specified in labeling | 1.9 X 10^-4 ΔA/min per U/L |
| Accuracy (n=150) | Not specified in labeling | Corr. Coefficient (r): 0.988 |
| Lin. Regression | Not specified in labeling | 0.44x + 0 U/L vs. COBAS INTEGRA α-Amylase (granulate) |
Cholesterol (CHOLL)
| Performance Characteristic | Acceptance Criteria (Predicate: COBAS INTEGRA Cholesterol (Granulate)) | Reported Device Performance (COBAS INTEGRA Cholesterol (Liquid)) |
|---|---|---|
| Assay Range | 0 - 20.7 mmol/L (0-8000 mg/dL with post dilution) | 0 - 18.1 mmol/L (0-7000 mg/dL with post dilution) |
| Precision (Level 1) | Mean: 5.0 mmol/L; %CV (w/r): 1.3; %CV (total): 1.1 | Mean: 5.3 mmol/L; %CV (w/r): 1.3; %CV (total): 2.2 |
| Precision (Level 2) | Mean: 6.3 mmol/L; %CV (w/r): 1.0; %CV (total): 1.2 | Mean: 6.7 mmol/L; %CV (w/r): 1.1; %CV (total): 2.5 |
| Precision (Level 3) | Mean: 8.0 mmol/L; %CV (w/r): 2.0; %CV (total): 1.5 | N/A |
| Sensitivity | 6.4 X 10^-2 ΔA per mmol/L | 8.8 X 10^-2 ΔA per mmol/L |
| Accuracy (n=214) | Corr. Coefficient (r): 0.995 | Corr. Coefficient (r): 0.998 |
| Lin. Regression | 1.04x + 0.1 mmol/L | 0.99x + 0.0 mmol/L vs. COBAS INTEGRA Cholesterol (granulate) |
HDL-Cholesterol Application (HDLL)
| Performance Characteristic | Acceptance Criteria (Predicate: COBAS INTEGRA HDL - Cholesterol Application (granulate)) | Reported Device Performance (COBAS INTEGRA HDL - Cholesterol Application (liquid)) |
|---|---|---|
| Assay Range | 0 - 5.0 mmol/L (0-193 mg/dL) | 0 - 5.0 mmol/L (0-193 mg/dL) |
| Precision (Level 1) | Mean: 0.82 mmol/L; %CV (w/r): 1.2; %CV (total): 2.7 | Mean: 0.20 mmol/L; %CV (w/r): 1.51.3*; %CV (total): 3.0 |
| Precision (Level 2) | Mean: 1.42 mmol/L; %CV (w/r): 0.85; %CV (total): 5.5 | Mean: 1.91 mmol/L; %CV (w/r): 0.26; %CV (total): 1.6 |
| Sensitivity | 6.4 X 10^-2 ΔA per mmol/L | 8.8 X 10^-2 ΔA per mmol/L |
| Accuracy (n=240) | Corr. Coefficient (r): 0.998 | Corr. Coefficient (r): 0.999 |
| Lin. Regression | 0.99x - 0.05 mmol/L | 0.99x + 0.03 mmol/L vs. COBAS INTEGRA HDL - Cholesterol Application (granulate) |
Note: There seems to be a typo for %CV (within run) in Level 1 of the HDLL liquid reagent (1.51.3).
Gamma-Glutamyltransferase (GGTL)
| Performance Characteristic | Acceptance Criteria (Predicate: COBAS INTEGRA GGT (Granulate)) | Reported Device Performance (COBAS INTEGRA GGTL (Liquid)) |
|---|---|---|
| Assay Range | 0 - 700 U/L (0-7000 U/L with post dilution) | 0 - 600 U/L (0-6000 U/L with post dilution) |
| Precision (Level 1) | Mean: 37.9 U/L; %CV (w/r): 0.67; %CV (total): 1.2 | Mean: 21 U/L; %CV (w/r): 0.83; %CV (total): 2.8 |
| Precision (Level 2) | Mean: 345 U/L; %CV (w/r): 0.46; %CV (total): 1.4 | Mean: 428 U/L; %CV (w/r): 0.54; %CV (total): 1.5 |
| Sensitivity | 5.0 X 10^4 ΔA/min per U/L | 6.8 X 10^4 ΔA/min per U/L |
| Accuracy (n=196) | Corr. Coefficient (r): 0.998 | Corr. Coefficient (r): 0.999 |
| Lin. Regression | 1.00x + 0 U/L | 1.00x - 1.2 U/L vs. COBAS INTEGRA GGTL (granulate) |
Glucose (GLUCL) - Serum and Plasma
| Performance Characteristic | Acceptance Criteria (Predicate: COBAS INTEGRA Glucose (Granulate)) | Reported Device Performance (COBAS INTEGRA Glucose (Liquid)) |
|---|---|---|
| Assay Range | 0 - 40 mmol/L (0-400 mmol/L with post dilution) | 0 - 40 mmol/L (0-400 mmol/L with post dilution) |
| Precision (Level 1) | Mean: 5.6 mmol/L; %CV (w/r): 1.2; %CV (total): 1.1 | Mean: 5.3 mmol/L; %CV (w/r): 1.7; %CV (total): 2.6 |
| Precision (Level 2) | Mean: 19.7 mmol/L; %CV (w/r): 0.97; %CV (total): 0.89 | Mean: 33.2 mmol/L; %CV (w/r): 0.72; %CV (total): 1.5 |
| Sensitivity | 9.3 X 10^-2 ΔA per mmol/L | 5.4 X 10^-2 ΔA per mmol/L |
| Accuracy (n=220) | Corr. Coefficient (r): 0.997 | Corr. Coefficient (r): 0.999 |
| Lin. Regression | 0.98x + 0.1 mmol/L | 1.05x - 0.2 mmol/L vs. COBAS INTEGRA Glucose (granulate) |
Glucose (GLUCL) - Urine
| Performance Characteristic | Acceptance Criteria (Predicate: COBAS INTEGRA Glucose (Granulate)) | Reported Device Performance (COBAS INTEGRA Glucose (Liquid)) |
|---|---|---|
| Assay Range | 0 - 16 mmol/L (0-160 mmol/L with post dilution) | 0 - 40 mmol/L (0-400 mmol/L with post dilution) |
| Precision (Level 1) | Mean: 0.27 mmol/L; %CV (w/r): 2.0 | Mean: 1.7 mmol/L; %CV (w/r): 1.7; %CV (total): 4.3 |
| Precision (Level 2) | Mean: 0.48 mmol/L; %CV (w/r): 0.99 | Mean: 37.1 mmol/L; %CV (w/r): 1.8; %CV (total): 2.9 |
| Sensitivity | 2.2 X 10^1 ΔA per mmol/L | 5.4 X 10^2 ΔA per mmol/L |
| Accuracy (n=120) | Not specified in labeling | Corr. Coefficient (r): 0.999 |
| Lin. Regression | Not specified in labeling | 1.01x - 0.02 mmol/L vs. COBAS INTEGRA Glucose (granulate) |
Glucose (GLUCL) - CSF
| Performance Characteristic | Acceptance Criteria (Predicate: COBAS INTEGRA Glucose (Granulate)) | Reported Device Performance (COBAS INTEGRA Glucose (Liquid)) |
|---|---|---|
| Assay Range | 0 - 20 mmol/L (0-360 mmol/L with post dilution) | 0 - 40 mmol/L (0-400 mmol/L with post dilution) |
| Precision (Level 1) | Mean: 4.7 mmol/L; %CV (w/r): 0.57 | Mean: 1.7 mmol/L; %CV (w/r): 1.6; %CV (total): 2.3 |
| Precision (Level 2) | Mean: 10.3 mmol/L; %CV (w/r): 0.23 | Mean: 3.3 mmol/L; %CV (w/r): 1.8; %CV (total): 1.9 |
| Sensitivity | 1.8 X 10^1 ΔA per mmol/L | 5.4 X 10^2 ΔA per mmol/L |
| Accuracy (n=212) | Not specified in labeling | Corr. Coefficient (r): 0.999 |
| Lin. Regression | Not specified in labeling | 1.02x - 0.17 mmol/L vs. COBAS INTEGRA Glucose (granulate) |
Lipase (LIPL)
| Performance Characteristic | Acceptance Criteria (Predicate: COBAS INTEGRA Lipase (Granulate)) | Reported Device Performance (COBAS INTEGRA Lipase (Liquid)) |
|---|---|---|
| Assay Range | 0 - 700 U/L (0-3500 U/L with post dilution) | 0 - 600 U/L (0-3000 U/L with post dilution) |
| Precision (Level 1) | Mean: 116 U/L; %CV (w/r): 1.7; %CV (total): 4.6 | Mean: 126 U/L; %CV (w/r): 1.9; %CV (total): 3.1 |
| Precision (Level 2) | Mean: 550 U/L; %CV (w/r): 2.1; %CV (total): 3.7 | Mean: 515 U/L; %CV (w/r): 1.3; %CV (total): 2.9 |
| Sensitivity | 5.6 X 10^-5 ΔA/min per U/L | 6.4 X 10^-5 ΔA/min per U/L |
| Accuracy (n=198) | Corr. Coefficient (r): 0.976 | Corr. Coefficient (r): 0.976 |
| Lin. Regression | 1.06x - 7 U/L | 0.82x + 16 U/L vs. COBAS INTEGRA Lipase (granulate) |
Urea/BUN (UREAL) - Serum and Plasma
| Performance Characteristic | Acceptance Criteria (Predicate: COBAS INTEGRA Urea/BUN (Granulate)) | Reported Device Performance (COBAS INTEGRA Urea/BUN (Liquid)) |
|---|---|---|
| Assay Range | 0 - 55 mmol/L (0-550 mmol/L with post dilution) | 0 - 40 mmol/L (0-400 mmol/L with post dilution) |
| Precision (Level 1) | Mean: 6.9 mmol/L; %CV (w/r): 0.85; %CV (total): 2.0 | Mean: 4.1 mmol/L; %CV (w/r): 2.3; %CV (total): 3.9 |
| Precision (Level 2) | Mean: 19.9 mmol/L; %CV (w/r): 1.0; %CV (total): 2.3 | Mean: 31.0 mmol/L; %CV (w/r): 0.89; %CV (total): 2.8 |
| Sensitivity | 6.8 X 10^-3 ΔA/min per mmol/L | 2.2 X 10^-2 ΔA/min per mmol/L |
| Accuracy (n=236) | Corr. Coefficient (r): 0.999 | Corr. Coefficient (r): 0.999 |
| Lin. Regression | 1.01x + 0.30 mmol/L | 1.00x + 0.1 mmol/L vs. COBAS INTEGRA Urea/BUN (granulate) |
Urea/BUN (UREAL) - Urine
| Performance Characteristic | Acceptance Criteria (Predicate: COBAS INTEGRA Urea/BUN (Granulate)) | Reported Device Performance (COBAS INTEGRA Urea/BUN (Liquid)) |
|---|---|---|
| Assay Range | 0 - 2200 mmol/L (0-5500 mmol/L with post dilution) | 0 - 2000 mmol/L (0-6000 mmol/L with post dilution) |
| Precision (Level 1) | Mean: 73 mmol/L; %CV (w/r): 0.99 | Mean: 421 mmol/L; %CV (w/r): 1.3; %CV (total): 1.8 |
| Precision (Level 2) | Mean: 345 mmol/L; %CV (w/r): 0.6 | Mean: 679 mmol/L; %CV (w/r): 1.2; %CV (total): 1.8 |
| Sensitivity | Not specified in labeling | 2.0 X 10^-2 ΔA/min per mmol/L |
| Accuracy (n=120) | Not specified in labeling | Corr. Coefficient (r): 0.999 |
| Lin. Regression | Not specified in labeling | 1.0X + 1.3 mmol/L vs. COBAS INTEGRA Urea/BUN (granulate) |
2. Sample sizes used for the test set and data provenance
The sample sizes for accuracy/correlation studies are provided in the tables above under "Sample size (n)". These values range from 114 to 240 for quantitative assays and 39 (positive) for LSD.
The data provenance is not explicitly stated as "country of origin" or "retrospective/prospective." However, given the context of a 510(k) submission for in vitro diagnostic reagents by Roche Diagnostic Systems, Inc. (located in Somerville, New Jersey, USA), it is highly likely these studies were conducted in a clinical laboratory setting in the USA. The studies are presented as direct comparisons between the new liquid reagents and existing granulate reagents or other legally marketed devices, implying they are prospective comparative studies assessing analytical performance.
3. Number of experts used to establish the ground truth for the test set and qualifications of those experts
For these chemical assays (e.g., ammonia, creatinine, cholesterol) and therapeutic drug monitoring (digitoxin), "ground truth" is typically established by the quantitative results of the predicate device or a reference method. The document does not mention "experts" in the sense of human readers adjudicating results, as these are quantitative in vitro diagnostic tests. The ground truth is the measured concentration or activity of the analyte as determined by the accepted reference method or predicate device functionality.
For LSD, which involves qualitative detection, the "ground truth" against which the COBAS INTEGRA LSD was compared appears to be GC/MS (Gas Chromatography/Mass Spectrometry), which is a gold standard analytical method for drug detection. The document does not specify the qualifications of individuals performing these GC/MS analyses or interpreting the results, but they would be trained laboratory personnel.
4. Adjudication method (e.g., 2+1, 3+1, none) for the test set
No adjudication method is mentioned. For these types of quantitative and qualitative analytical tests, "adjudication" by experts in the context of diagnostic imaging or pathology interpretation is not applicable. The comparison is based on numerical results compared to an established method or predicate.
5. If a multi-reader multi-case (MRMC) comparative effectiveness study was done, if so, what was the effect size of how much human readers improve with AI vs without AI assistance
No MRMC study was conducted. This type of study is relevant for medical imaging or pathology devices where human interpretation is a key component, often assisted by AI. The submitted devices are reagents for automated clinical analyzers, where the output is a numerical value or a qualitative positive/negative result, not an image requiring human interpretation. Therefore, there's no mention of AI assistance or human reader improvement.
6. If a standalone (i.e., algorithm only without human-in-the-loop performance) was done
Yes, the studies presented are effectively standalone performance evaluations of the reagent-analyzer system. These are not "algorithm-only" studies in the modern AI sense, but rather a direct assessment of the analytical performance of the new liquid reagent format on the COBAS INTEGRA Analyzer. The results (e.g., assay range, precision, accuracy) reflect the performance of the integrated system without direct human-in-the-loop interpretation impacting the primary measurement.
7. The type of ground truth used (expert consensus, pathology, outcomes data, etc.)
The type of ground truth used varies slightly based on the assay, but generally involves:
- Quantitative Assays (Ammonia, Creatinine, Cholesterol, HDL-Cholesterol, GGT, Glucose, Lipase, Urea/BUN): The ground truth for these assays is the quantitative result obtained from the predicate device method. The accuracy is assessed by correlating the results of the new liquid reagents with the predicate (often the granulate version of the same Roche COBAS INTEGRA reagents or another established method like Abbott TDx/TDxFLx for Digitoxin). Linear regression analysis is used to demonstrate agreement.
- Qualitative Assay (LSD): The ground truth for LSD detection is established by a more definitive analytical method, specifically Gas Chromatography/Mass Spectrometry (GC/MS).
8. The sample size for the training set
The document does not explicitly delineate a "training set" in the context of machine learning or AI development. For these chemical assays, the development of the reagents and their formulation would involve extensive R&D and optimization, which could be considered an iterative development process, but it's not described as a distinct "training set" with separate ground truth establishment. The data presented in the tables are for validation or verification of the final product.
9. How the ground truth for the training set was established
As there is no explicitly defined "training set" in the submitted documentation related to AI/ML, there is no description of how ground truth for such a set was established. The development of reagents relies on established chemical and biochemical principles, and performance characteristics are determined through standard analytical validation procedures using reference materials and comparative studies against predicate methods.
§ 862.1070 Amylase test system.
(a)
Identification. An amylase test system is a device intended to measure the activity of the enzyme amylase in serum and urine. Amylase measurements are used primarily for the diagnosis and treatment of pancreatitis (inflammation of the pancreas).(b)
Classification. Class II.