Reply to Letter by Seibold regarding Monika Reddy,Narvada Jugnee,Sinthuka Anantharaja,and Nick Oliver,Switching from Flash Glucose Monitoring to Continuous Glucose Monitoring on Hypoglycemia in Adults with Type 1 Diabetes at High Hypoglycemia Risk: The Extension Phase of the I HART CGM Study

Dear Editor:

We thank Seibold for his comments ¹ on the extension phase of our randomized study of flash and continuous glucose monitoring (CGM) in adults with type 1 diabetes at highest risk of hypoglycemia (the IHART CGM study). ²

We have previously responded to the comments by Seibold and Schlaeger on the original article presenting the data from the initial 8-week randomized study. ^{3 –5}

The IHART CGM study protocol allowed participants to use both the Freestyle Libre and the Dexcom G5 according to the manufacturers' user guides. We, therefore, did not request participants to conduct routine capillary finger-prick testing to minimize the potential additional burden for participants. However, we acknowledge the methodological challenges with comparing outcomes derived from different devices with reported differences in sensitivity and specificity at different glucose ranges ^6,7 and recognize the importance of a standard reference methodology when comparing different interstitial glucose monitoring technologies in clinical trials. ⁸

We have already discussed the sample size as a limitation in our article. ² Nevertheless, the IHART CGM study is the only published head-to-head randomized, controlled trial comparing flash and CGM to date. The small study size may well have contributed to the wide distribution of the original baseline data for percentage time <3.0 mmol/L, but a comparison of all glucose outcomes, derived from the run-in blinded CGM data, between the CGM and flash glucose monitoring groups was performed (using a nonparametric test) and showed no statistical difference between groups at baseline. For example, the percentage time in hypoglycemia (<3.0 mmol/L) for the flash group at baseline versus CGM was 4.7 (3.2–7.0) and 3.1 (1.0–11.7), respectively, P = 0.5. The baseline Hypoglycaemia Fear Survey (HSF-II) worry subscore at randomization for the flash and CGM groups was 27.5 (18.0–34.5) and 40.5 (24.0–52.5), respectively, P = 0.13. There was, therefore, no demonstrable baseline imbalance or bias.

Seibold is correct that some secondary outcomes did not differ between the groups in IHART CGM but that is true of the majority of clinical studies, including the IMPACT study, which showed that, with FreeStyle Libre, HbA1c did not objectively fall compared with self-monitored blood glucose. ⁹ Primary and secondary outcomes are carefully selected to align with the study design and should not be reprioritized after analysis.

We agree that both flash glucose monitoring and real-time CGM technologies have potential utility for people with type 1 diabetes using a multiple dose insulin injection regimen. However, we believe that people with type 1 diabetes and impaired awareness of hypoglycemia should be offered CGM with real-time alerts and alarms as first line technology intervention, and that this is supported by our study results.