Abstract

Aims To investigate associations between HbA1c variability and all-cause mortality in individuals with diabetes, accounting for average HbA1c level Methods Mean HbA1c and variability score (HVS) were estimated for people aged 31–90 with diabetes (type 1 = 20,347, type 2 = 409,821) with 4 + HbA1c measurements recorded in the Clinical Practice Research Datalink in 2011–14 and alive on 1/1/2015. Cox models estimated hazard ratios (HR) for all-cause mortality, ascertained from national linked mortality data during 2015–17. HbA1c level and variability were mutually adjusted for each other and other measured confounders. Results Greater HbA1c variability was associated with younger age, non-white ethnicities (type 1 only), obesity, co-morbidities, and living in deprived areas. During follow-up, 1,043 (5.1 %) individuals with type 1 diabetes and 40,723 (9.9 %) individuals with type 2 diabetes died. In those with the most HbA1c variability compared to the least (HVS = 80–100 vs 0–20), the estimated adjusted HRs for mortality were 2.78(95 %CI 2.15, 3.60) in type 1 diabetes and 1.91(1.83, 1.99) in type 2 diabetes. Conclusions Variability in HbA1c was associated with greater subsequent mortality among people living with diabetes, independent from average HbA1c. Future research should investigate whether reducing HbA1c variability over time in selected patients lowers mortality risk independent of HbA1c level improvements.