Implementation of flash glucose monitoring in four pediatric diabetes clinics: controlled before and after study to produce real-world evidence of patient benefit

Aims: To assess the real-world evidence for flash glucose monitoring (Abbott FreeStyle Libre) for children with type 1 diabetes in terms of glucose control, secondary healthcare resources and costs.

Research design and methods: We conducted a controlled before and after study (approximately 12 months before and after) using routinely collected health record data on children who start using flash monitors and a control population of children with self-monitoring of blood glucose (SMBG). Our population-based sample of eligible individuals using flash monitoring (n=114) and controls (n=80) aged between 4 and 18 years was drawn from four paediatric diabetes clinics (secondary care) in the South West England. Outcome measures included: glycated hemoglobin (HbA1c), frequency of BG tests; frequency of sensor scans; time in recommended glucose range; short-term complications (hypoglycemia, diabetic ketoacidosis and related illness resulting in investigation) and secondary care costs.

Results: After adjustment for age, time since diagnosis, deprivation and the test modality (point of care or laboratory), the mean HbA1c reading for controls was 61.2 (mmol/mol) for the period before and 63.9 after. For individuals using flash monitoring, the adjusted mean HbA1c reading was 64.6 for the period before implementation and 63.8 after. Rates of short-term complications were low across all groups in the study. Whereas the 'after' flash monitoring group had substantially higher incremental costs (+£703 vs the flash monitoring 'before' comparison and +£841 vs contemporaneous SMBG controls), these cost differences were driven by primary care prescribing (sensor costs).

Conclusions: There was some indication that flash monitoring might help young people improve the control of their diabetes but for our sample, the difference between finger-prick testing and flash monitoring was not clinically significant (HbA1c improvement <5 mmol/mol). Given the pace of technological change within diabetes, research efforts should now facilitate the real-time analysis of long-term routine data on flash and continuous glucose monitors.