Accuracy Evaluation of a Blood Glucose Monitoring System According to EN ISO 15197:2015

The accuracy of the blood glucose readings provided by the GLUCOCARD SM blood glucose meter (A. Menarini Diagnostics, Florence, Italy) was assessed according to the guidelines (EN ISO 15197:2015, paragraph 6.3) introduced to establish acceptable performance for blood glucose monitoring systems. ¹ This is a new blood glucose meter recently introduced into the European market, offering advanced features like Bluetooth Low Energy on board, allowing patients to have glucose values automatically sent to a smartphone.

The study was performed within the Joint Clinical Research Facility, Swansea University in compliance with Good Clinical Practice (GCP) and approved by the relevant ethics committee. Subjects (N = 100; male and female) who met the inclusion criteria had a fingerstick capillary blood sample taken in appropriate anticoagulant. This sample was used for measurement of the blood glucose level in the meter (3 lots of strips in duplicate), and a laboratory reference method analyzer, YSI 2300.

Plasma glucose was analyzed in duplicate using a YSI 2300 Stat Plus (Yellow Springs Instruments, Fleet, UK) and blood glucose determined on 6 different meters using 3 lots of test strips, which were all provided by the manufacturer. For YSI measurement, blood samples were collected into capillary collection tubes (Lithium Heparin Microvette, Sarstedt, Leicester, UK) and centrifuged prior to plasma glucose measurement. In addition, hematocrit was measured using a HemoControl analyzer (EKF, Cardiff, UK). Daily control measures were carried out on all meters and instruments. Other than in the modified samples, all blood sampling and analysis were performed within a 5-minute time frame. To meet the extreme glucose concentration criterion, samples aimed to be <80 mg/dL (4.44 mmol/L) were incubated to allow glucose to hydrolyze (glycolysis). Similarly, to obtain samples >300 mg/dL (16.65 mmol/L) a supplement of glucose was added. A maximum of 22 blood samples were modified in this way in accordance with EN ISO 15197:2015. Modified samples had pO₂ concentrations equivalent to capillary blood prior to assay.

Good agreement and linearity were observed between the blood glucose results from the 3 different lots and the YSI reference results (Fig 1). Combined data from the 3 lots showed 99% of readings from the blood glucose meter met each of the individual criteria. At glucose concentration <100 mg/dL, 168/174 (95% [lot 1], 97% [lot 2], 98% [lot 3]) results fell within ±15 mg/dL of the YSI values. At glucose concentrations ≥100 mg/dL, 423/426 (100% [lot 1], 99% [lot 2], 99% [lot 3]) results fell within ±15% specified by EN ISO 15197:2015. The overall combined results demonstrated conformance (591/600 [99%] within ±15 mg/dL and within ±15%). As shown in Figure 1B, a total of 99.8% values fell within zone A of the consensus error grid, ² defined as “clinically accurate measurement,” and the remaining 0.2% fell in zone B demonstrating compliance with criterion B specified by EN ISO 15197:2015.

Figure 1.

(A) System accuracy plot for meter vs YSI 2300 plasma glucose concentration. Data are shown from all 3 test strip lots (each lot plotted with a different marker). The area inside the solid lines represents the minimum acceptable accuracy from EN ISO 15197:2015. (B) Consensus error grid plot of meter vs YSI 2300 plasma glucose.

These results demonstrate good agreement between the test blood glucose meter and the reference laboratory method, exceeding the minimum acceptable system accuracy criteria specified by EN ISO 15197:2015.