Introduction: Glucose sensor usage is increasing in the paediatric type 1 diabetes population. The sensor downloads can provide valuable information about glycaemic levels over a 90-day period and generate an estimated HbA1c based on the average glucose level.
Aim: We aimed to test whether the sensor-estimated HbA1c over 90 days was an accurate prediction of the measured HbA1c and whether its accuracy correlated with percentage sensor data captured.
Methods: Over a 12-week period, 90-day sensor downloads were collected from children with type 1 diabetes who were wearing a glucose sensor (Freestyle Libre or Dexcom G5) on the day they were due their 3-monthly HbA1c laboratory test. The Freestyle Libre handset was downloaded in clinic and the Dexcom G5 was accessed through Clarity online portal to generate the reports. Each family provided informed consent for their data to be used in the study. The HbA1c was measured by ion-exchange high-performance liquid chromatography (HPLC) from EDTA whole blood. The difference between the measured and calculated HbA1c was calculated (delta HbA1c).
Results: Twenty four children who were wearing glucose sensors had HbA1c tests during the study period (20 were wearing Freestyle Libre and 4 Dexcom G5). The mean laboratory HbA1c was 7.85% (SD 1.39, Range 5.8 to 12%). The mean predicted HbA1c was 7.66% (SD 1.52, Range 5.3 to 11.4%). The mean delta hbA1c was 0.18% (SD 0.58, Range -1.1 to +1.4%), with a tendency for the prediction to be lower than the measured HbA1c in 67% of cases. The mean delta HbA1c for the Freestyle Libre was 0.1% and for the Dexcom G5 0.7%. The estimated HbA1c was within 0.5% of the laboratory HbA1c 50% of the time and within 0.75% 79.2% of the time. Bland Altman Analysis confirmed there was no relationship between the level of HbA1c and the delta hbA1c, or between percentage sensor data captured and delta HbA1c.
Conclusions: There is a tendency for estimated HbA1c to be lower than the measured HbA1c but the mean difference is small. The delta HbA1c is significant in a few individuals but there was no correlation with lower sensor wear time or higher HbA1c. The sensor download provides a useful estimate of the HbA1c and the estimated HbA1c is within 0.75% of the measured HbA1c 79.2% of the time. With increasing sensor accuracy the estimated HbA1c may eventually replace the need for a 3-monthly HbA1c blood test.
19 - 21 Sep 2019
European Society for Paediatric Endocrinology