Evaluation of a Methodology for Estimating HbA1c Value by a New Glucose Meter

Abstract

Background: Accuracy/robustness of HbA1c estimation (eA1c) with an algorithm built into the MyStar Extra blood glucose (BG) meter has been demonstrated by in silico testing. We evaluated the performance and use of eA1c in a clinical setting.

Methods: Subjects took the BG meter home for 4 months to obtain eA1c in this open-label, single-center study. Laboratory HbA1c values were obtained approximately every 2 weeks and the corresponding eA1c documented. Subjects completed a questionnaire at study end (NCT01885546).

Results: There were 133 enrolled subjects (mean [SD] age 60.0 [15.0] years, 69 males, 104 with diabetes, HbA1c 7.0% [1.4]). A total of 1008 pairs of eA1c and laboratory HbA1c values were available. In subjects with diabetes, 97.5% of the eA1c results fell within ± 20% of the laboratory HbA1c, 95.0% within ± 18%, and 90.7% within ± 15%. When results were limited to the reportable HbA1c range of ≥ 6 to ≤ 10%, 99.3% of eA1c values fell within ± 20% of the laboratory HbA1c, 98.5% within ± 18%, and 96.2% within ± 15% Most subjects agreed/strongly agreed that the eA1c section in the user guide and flash cards was easy to follow (72%), they would use the system to track their eA1c (70%), they found the eA1c tool helpful (79%), and the tool may motivate them to manage their diabetes better (83%).

Conclusions: Accuracy of the eA1c feature in this clinical setting was similar to the performance in silico. The majority of subjects found this tool helpful and agreed it may motivate to manage their diabetes better.

Keywords: MyStar Extra blood glucose meter; eA1c; eA1c questionnaire; estimated HbA1c; self-monitored blood glucose.

Conflict of interest statement

Declaration of Conflicting Interests: The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

© 2015 Diabetes Technology Society.

Figures

Figure 1.

Scatter plot of eA1c versus laboratory HbA1c for all subjects (A) and only subjects with diabetes (C) and histogram of the relative difference between laboratory HbA1c and eA1c for all subjects (B) and only subjects with diabetes (D). The data were stratified by laboratory HbA1c (vertical dashed lines) or eA1c (horizontal dashed lines) values below 6% and above 10%. The eA1c values > 10 are shown because internally the meter continued to calculate the eA1c values regardless of whether or not the value is 10. Therefore, for accuracy evaluation purposes the actual values are shown. In the field, the meter would show “low” or “high” to the users with values 10, respectively.

Figure 2.

eA1c questionnaire results.