(65 days)
The Abbott ARCHITECT™ Folate assay is a Chemiluminescent Microparticle Folate Binding Protein assay for the quantitative determination of folate in human serum, plasma, and red blood cells on the Abbott ARCHITECT™ i System. Measurements obtained by this device aid in the diagnosis and treatment of megaloblastic anemia.
The ARCHITECT Folate assay is a Chemiluminescent Microparticle Folate Binding Protein assay for the quantitative determination of folate in human serum, plasma, (tripotassium EDTA, lithium heparin, or sodium heparin) and red blood cells (tripotassium EDTA). The ARCHITECT Folate assay is calibrated with Abbott ARCHITECT Folate Calibrators. Abbott ARCHITECT Folate Controls are assayed for the verification of the accuracy and precision of the Abbott ARCHITECT™ i System.
Here's an analysis of the provided text regarding the Abbott ARCHITECT™ Folate assay, structured to address your specific questions about acceptance criteria and study details:
Acceptance Criteria and Study Details for Abbott ARCHITECT™ Folate
The Abbott ARCHITECT™ Folate assay is a Chemiluminescent Microparticle Folate Binding Protein assay for the quantitative determination of folate. The presented study aimed to demonstrate substantial equivalence to a legally marketed predicate device, the BioRad Quantaphase II Folate Radioassay. The primary method for demonstrating this was through correlation studies using patient samples.
1. Table of Acceptance Criteria and Reported Device Performance
The acceptance criteria are implied by the statistical measures of correlation and agreement between the new device and the predicate device. While explicit "acceptance criteria" are not numerically stated beforehand (e.g., "r > 0.95"), the reported statistics (correlation coefficient, slope, and intercept with 95% confidence intervals) are the performance measures that the FDA accepted as demonstrating substantial equivalence. The FDA letter confirms the device was found substantially equivalent.
| Performance Metric | Acceptance Criteria (Implied) | Reported Device Performance (Serum) | Reported Device Performance (Whole Blood) |
|---|---|---|---|
| Correlation Coefficient (r) | High correlation (e.g., typically >0.90 for substantial equivalence in IVD correlation studies) | 0.927 | 0.929 |
| Slope (Least Squares) | Close to 1 (indicating proportional agreement) | 0.97 (0.92, 1.01) | 1.00 (0.94, 1.06) |
| Intercept (Least Squares) | Close to 0 (indicating lack of constant bias) | 0.44 (0.02, 0.87) | 36.04 (4.62, 67.46) |
| Slope (Passing-Bablok) | Close to 1 | 1.08 (1.05, 1.12) | 1.14 (1.09, 1.20) |
| Intercept (Passing-Bablok) | Close to 0 | -0.18 (-0.47, 0.06) | -18.83 (-40.97, 3.59) |
Note on Interpretation:
- For the Least Squares regression, a slope of 1 and an intercept of 0 would indicate perfect agreement. The reported values for serum (slope 0.97, intercept 0.44) and whole blood (slope 1.00, intercept 36.04) are generally close to these ideals, especially considering their 95% confidence intervals.
- Passing-Bablok regression is often preferred for method comparison studies as it is robust against outliers and does not assume errors only in one method. The slopes of 1.08 and 1.14, and intercepts of -0.18 and -18.83, also demonstrate good agreement, though a slight proportional and constant bias can be inferred from the deviations from 1 and 0, respectively. However, given the high correlation and the FDA's acceptance, these were deemed acceptable.
2. Sample Size Used for the Test Set and Data Provenance
- Sample Sizes:
- Serum: n = 333
- Whole Blood (referred to as "red blood cells" in other parts of the document): n = 160
- Data Provenance: Not explicitly stated (e.g., country of origin). The study details do not specify if the data was retrospective or prospective. Given the nature of method comparison studies at the time, it's common for them to be prospective clinical samples collected for the purpose of the study.
3. Number of Experts Used to Establish Ground Truth for the Test Set and Qualifications
N/A - This device is an in vitro diagnostic (IVD) assay for quantitative determination of a biomarker (folate). "Ground truth" is established by the analytical result of the predicate device (BioRad Quantaphase II Folate Radioassay) on the same samples, not by expert interpretation. Therefore, experts are not used in this context to establish ground truth.
4. Adjudication Method for the Test Set
N/A - Adjudication methods (like 2+1, 3+1) are relevant for studies where subjective interpretation (e.g., by radiologists) is being evaluated and discrepancies need to be resolved. This is a method comparison study for an IVD, so "adjudication" in that sense is not applicable. The comparison involves direct quantitative measurements.
5. If a Multi Reader Multi Case (MRMC) Comparative Effectiveness Study was done
No, an MRMC comparative effectiveness study was not done. This study is an analytical method comparison for an in vitro diagnostic device, not a study evaluating human reader performance with or without AI assistance.
6. If a Standalone (i.e. algorithm only without human-in-the-loop performance) was done
Yes, in a sense, a "standalone" evaluation was performed. The ARCHITECT Folate assay (the "algorithm/device") directly measured folate levels in samples, and these results were compared to the standalone results of the predicate device. There is no human-in-the-loop component in the measurement process itself for either the new device or the predicate. The performance reported (correlation, slope, intercept) is the performance of the device itself.
7. The Type of Ground Truth Used
The "ground truth" for this method comparison study is the results obtained from the legally marketed predicate device, the BioRad Quantaphase II Folate Radioassay, on the same patient samples. This is a common approach for demonstrating substantial equivalence for new IVD devices.
8. The Sample Size for the Training Set
N/A - This is a traditional IVD method comparison study, not a machine learning model development. Therefore, there isn't a "training set" in the sense of data used to train an algorithm. The study described is entirely a validation or test set where the new device's analytical performance is compared against an established method.
9. How the Ground Truth for the Training Set Was Established
N/A - As there is no training set in the machine learning context, this question is not applicable.
§ 862.1295 Folic acid test system.
(a)
Identification. A folic acid test system is a device intended to measure the vitamin folic acid in plasma and serum. Folic acid measurements are used in the diagnosis and treatment of megaloblastic anemia, which is characterized by the presence of megaloblasts (an abnormal red blood cell series) in the bone marrow.(b)
Classification. Class II.