Point-of-Care Fingerstick PT/INR System

Against the point-of-care comparator (n = about 400), the INR slope was slightly below 1, the intercept near zero and the correlation above 0.9, with candidate results spanning 0.8 to 7.7 INR. Against the laboratory coagulation analyzer on venous plasma (n = about 400), the INR slope was about 1.1, with a small negative intercept and a correlation above 0.9. In seconds, the slope against the laboratory method was about 1.3, again with a correlation above 0.9.

Capillary repeatability from duplicate fingersticks by untrained operators ranged from about 4% to 5% CV across four INR intervals, from below 1.9 INR up to the 4.6 to 8.0 interval. Overall reproducibility across the three CLIA-waived sites was about 3% CV for the low control level (mean about 1.2 INR) and about 1% CV for the therapeutic level (mean about 2.6 INR). The single-site 20-day within-lab precision was about 1% to 3% CV across low, therapeutic and high control levels.

Hematocrit between 25% and 55% did not significantly affect results. The normal range was 0.9 to 1.1 INR, with more than 95% of healthy-subject results within it. FDA found the system substantially equivalent, and FDA's CLIA database lists it as waived for prothrombin time.

The two comparisons tell different stories. Against the capillary point-of-care comparator, the slope sat just below 1, indicating slightly lower readings at higher INRs. Against the laboratory method, a slope of about 1.1 with a small negative intercept means fingerstick INRs read proportionally higher than venous-plasma INRs as values increase. Because the two comparisons differ in matrix (capillary whole blood versus venous citrated plasma), they answer different questions and should be read together. The PT-seconds slope of about 1.3 illustrates why INR, not seconds, is the standardized unit for comparing different thromboplastin systems.

FDA's dual-submission guidance recommends bias at medical decision levels and EP21 total error for quantitative tests, but the public decision summary reports regression and precision results only, without numeric acceptance limits for the comparison. The 510(k) summary states pre-specified precision criteria of CV ≤5% for intermediate precision and CV ≤6% for the multi-site reproducibility study; all reported values fall within them. Note that precision varied more at the extremes. The 4.6 to 8.0 INR interval showed the highest repeatability CV (about 5%), which is where fewer subjects are available and where clinical action is most urgent.

The system determines INR for monitoring oral anticoagulation with warfarin (a vitamin K antagonist) in fresh capillary whole blood from a fingerstick, and reports results both in INR and in seconds. It is intended for patients 18 years or older who have been stable on vitamin K antagonist therapy for at least six weeks, and not for patients transitioning from heparin. It is a prescription device for multi-patient use in professional healthcare settings, including CLIA-waived and point-of-care settings.

Under a dual submission, the clinical evidence had to come from intended-use conditions. That meant untrained operators at CLIA-waived point-of-care sites collecting and testing fingerstick samples themselves, from both warfarin patients and healthy subjects, across a 0.8 to 8.0 INR (9.6 to 96.0 seconds) reportable range. Those results needed to agree with a cleared point-of-care PT/INR comparator and with a laboratory coagulation analyzer on venous plasma collected in 3% citrate.

Precision also had to be shown in the hands of untrained operators at waived sites, not only in a development laboratory. FDA's decision summary states that the reportable range itself was established through the method comparison studies, so the enrolled population had to cover the full claimed range.

Prothrombin time tests are Class II devices under 21 CFR 864.7750. The CFR does not codify device-specific special controls for this regulation. The precision, method comparison, interference and coagulation-specific expectations seen in decision summaries are anchored in FDA-recognized CLSI standards. This submission cited EP05-A3, EP09c, EP07, EP14-A3, H47-A2 (one-stage PT and APTT testing), GP41 (venous specimen collection) and POCT14 (point-of-care coagulation testing and anticoagulation monitoring).

A CLIA waiver requires a separate showing that a test is simple and has an insignificant risk of an erroneous result. FDA's 2020 CLIA waiver guidance asks for comparison studies at a minimum of three sites representative of waived settings, with 1 to 3 untrained operators per site and at least nine untrained operators overall, using only the instructions intended for those users. Under the Dual 510(k) and CLIA Waiver by Application pathway, also described in 2020 guidance, a single set of comparison and reproducibility studies run by untrained operators can support both substantial equivalence and the waiver. For quantitative tests, that guidance recommends regression analysis with bias estimated at medical decision levels and at the limits of the measuring interval, total error estimated per CLSI EP21, and reproducibility at a minimum of three of the comparison sites with the same number of untrained operators (likely 2 or 3) at each.

Point-of-care INR testing combines several difficulties. A capillary sample has to be compared with venous citrated plasma, and the clot-time result is converted to INR through lot-specific calibration traceable to the WHO reference thromboplastin. Warfarin dosing hinges on small INR differences. A reportable range extending to 8.0 INR also means the study has to enroll and test patients with markedly elevated INRs, not just therapeutic ones.

• Having untrained operators perform both the fingerstick collection and the test reflects how the device is used at waived sites, where specimen collection technique is a major source of error. This matches FDA's recommendation that dual-submission studies use operators with the least training likely to be encountered.
• Using 3–5 CLIA-waived sites and about 10 untrained operators met or exceeded the minimums in FDA's CLIA waiver guidance (at least three sites, at least nine untrained operators, 1 to 3 per site).
• Two comparators answered two questions. The cleared point-of-care device tested equivalence on the same capillary matrix, and the laboratory analyzer on citrated venous plasma tested accuracy against the method used in routine anticoagulation management.
• Enrolling a large warfarin cohort alongside healthy subjects populated the range from normal through the high INRs needed to support an upper limit of 8.0 INR; about 100 repeatability results fell in the 4.6 to 8.0 INR interval.
• Splitting precision into duplicate-fingerstick repeatability in patients and control-material reproducibility across waived sites separated sampling variability from instrument, lot and operator variability. This follows the guidance's suggestion to explore small-scale repeatability with intended-use samples when specimens such as capillary blood are not stable.