Page 39 - Glucose Monitoring Devices

The process for premarket approval of SMBG devices 35

and conditions (e.g., ketoacidosis) and an evaluation of the stability of reagents and
materials, but these changes only minimally affect the analytical performance of
BGMs.
Precision
This term represents how closely and consistently repeated measurements per-
formed by SMBG on a given sample align with each other. According to the latest
ISO 15197 2013 guidelines, testing must demonstrate repeatability on at least five
venous blood samples with defined glucose concentrations in the hypoglycemic,
euglycemic, and hyperglycemic ranges. These guidelines also assess intermediate
measurement precision by requiring daily measurements over 10 days.

Interference evaluation
The 2013 update to ISO 15197 introduced the evaluation of performance in the
presence of various hematocrit ranges and other possible interfering substances
that might be found in the blood such as medications or physiologically occurring
substances. For hematocrit, the BGM must be tested in five different hematocrit
ranges at each of three defined glucose concentrations. Other interfering substances
must be tested for at least two defined glucose concentrations. In each of these
different testing environments, the ISO 15197:2013 standard allows for up to: 1)
10 mg/dL difference between the test sample and control sample for glucose
concentrations 100 mg/dL and 2) 10% difference for glucose concentrations
>100 mg/dL.

Accuracy
Accuracy refers to how closely the BGM system’s measurements align with refer-
ence values obtained by a gold standard technique. ISO 15197:2013 requires that
at least 95% of measurements fall within 15 mg/dL of the reference value for
blood glucose (BG) concentrations <100 mg/dL and within 15% at BG
concentrations 100 mg/dL. The previous 2003 version was less rigorous,
requiring 15 mg/dL of reference for BG concentrations <75 mg/dL and 20%
for concentrations 75 mg/dL.
In addition, the 2013 update introduced the requirement that at least 99% of
measurement results fall within the consensus error grid (CEG) zones A and B
[12]. The CEG has five zones, characterized by different clinical risks to the patient.
By emphasizing that 99% of results fall in zones A (no effect on clinical action) and
B (little or no effect on the clinical outcome), this addition reduces the maximum
amount of clinically unacceptable results to 1%, along with the maximum amount
of analytically unacceptable results to 5%.
Furthermore, the 2013 update increased the minimum number of test strips lots
being evaluated for accuracy from one to three, allowing for a more comprehensive
assessment to compensate for lot-to-lot variability in performance, which has been
shown to significantly affect the accuracy of a system [13].