Change in glycated haemoglobin in adults with type 2 diabetes on basal-bolus insulin regimens following commencement of Freestyle Libre use

BACKGROUND
Most of the evidence supporting the efficacy of Freestyle Libre has accrued in people with type 1 diabetes. A randomised controlled trial of Freestyle Libre in people with type 2 diabetes on intensive insulin therapy did not demonstrate any significant change in HbA1c although hypoglycaemia was significantly reduced1. More recently, real-world evidence has suggested Freestyle Libre is associated with clinically relevant reductions in HbA1c in those with elevated HbA1c at baseline using intensive insulin therapy2,3. A number of studies have also suggested a potential benefit for CGM in the non-insulin treated type 2 diabetes4.Follow up duration in the existing literature is relatively short (up to 6 months) so we sought to assess longer-term outcomes in people with type 2 diabetes.
METHODS
Observational study of adults with type 2 diabetes attending a single Scottish diabetes centre. As a service evaluation of routinely collected data this project did not require ethical approval. All adults with type 2 diabetes on multiple daily injection insulin regimens were eligible for funded Freestyle Libre in our centre. Individuals were included if: (1) established on a basal bolus insulin regimen > 12 months and (2) HbA1c measured in the 12 months preceding and the 3 – 12 months following commencement. 770 people with type 2 diabetes, on basal bolus regimens, were identified as Freestyle Libre users, however paired HbA1c data were available in 304 (40%). For logistic regression analysis, a large cohort of people with type 1 diabetes were added (n = 987) to determine whether diabetes type predicted glycaemic response. Results are presented as median (IQR). Paired data were compared with Wilcoxon signed rank tests and unpaired data with Wilcoxon rank-sum test. Correlations were assessed by Spearman correlation coefficient. Categorical data were compared by χ2 test. P values <0.05 were considered statistically significant. Statistical analyses were performed using R Studio (version 2023.12.1).
RESULTS
57% of individuals were male (173/304). Median age was 61 years (IQR 52 – 69) and diabetes duration was 18 years (12 – 23). Median HbA1c at baseline was 73 mmol/mol (60 – 87) and BMI was 31.6 kg/m2 (27.6 – 36.3). 43% initially commenced the first-generation Freestyle Libre with the remainder commencing Freestyle Libre 2; this was not associated with any difference in outcomes.
Changes in HbA1c for the total cohort and stratified by baseline HbA1c are presented in table 1. Overall, a small reduction in HbA1c was noted across 3 years of follow up, with larger sustained reduction in HbA1c observed in those with highest baseline HbA1c and a small rise in HbA1c in those with lowest HbA1c at baseline. Higher HbA1c at baseline was correlated with greater reduction in HbA1c at 3 – 12 months (R -0.514, p < 0.001) (figure 1). 124/304 (41%) experienced a >5 mmol/mol reduction in HbA1c at 3 – 12 months; this was likelier in younger individuals (58 years [49 – 66] vs. 62 [54 – 70], p = 0.002) and those with higher HbA1c at baseline (84 mmol/mol [73 – 97] vs. 66 [55 – 76], p <0.001) but was not associated with sex (40% of women vs 42% of men, p = 0.735), diabetes duration (17 years [11 – 22] vs. 18 [12 – 24], p = 0.231), BMI (31.6 kg/m2 [28.0 – 35.1] vs. 31.5 [27.4 – 36.7], p = 0.900) or socioeconomic deprivation quintile (39% in the most deprived quintile vs. 42% in the least deprived quintile, p = 0.841). In logistic regression analysis, HbA1c response > 5mmol/mol (at 3 – 12 months) was only independently associated with HbA1c at baseline (OR 1.06 per mmol/mol [95% CI 1.046 – 1.063], p <0.001) but not with age (OR 0.994 per year [0.987 – 1.002], p = 0.135), BMI (OR 1.014 per kg/m2 [0.995 – 1.036], p = 0.149) or diabetes type (OR for type 1 diabetes vs. type 2 0.998 [0.726 – 1.375], p = 0.989).
CONCLUSIONS
These data suggest that those with highest HbA1c at baseline have a clinically relevant reduction in HbA1c following commencement of Libre. The durability of these results, over a three-year period, suggest this is a real phenomenon rather than regression to the mean in people with high baseline HbA1c. Importantly, these changes are entirely consistent with the magnitude of change observed in type 1 diabetes. We were unable to collect data on hypoglycaemia and quality of life which are other important clinical metrics which are potentially influenced by Freestyle Libre use1. Indeed, the small rise in HbA1c in people with lower HbA1c at baseline may reflect therapy changes to avoid hypoglycaemia. We cannot exclude the possibility of HbA1c reduction by additional therapy changes, although these people were long established on basal bolus insulin regimens and the introduction of other glucose lowering agents, at this stage, is not typical. These data, when considered in the context of previously published data, support the use of Freestyle Libre in people with type 2 diabetes using basal bolus insulin regimens. It will be of interest to determine whether a wider role exists for Freestyle Libre in those on twice-daily, basal-only insulin regimens and non-insulin treated type 2 diabetes. To date, only relatively small studies have assessed these cohorts, suggesting an urgent need for larger, clinically representative randomised controlled trials.