Oral Anticoagulants and risk of Type 2 Diabetes among adults with atrial
fibrillation in the United Kingdom
Abstract
Background: A number of observational studies and small trials have
suggested that higher levels of circulating vitamin K1 might improve
insulin sensitivity and therefore protect against the development of
Type II Diabetes (T2DM). The use of vitamin K-inhibitors, such as
warfarin, may lead to an increased risk of developing T2DM compared to
alternative oral anticoagulants (OACs). Objective: To compare the risk
of developing T2DM among atrial fibrillation patients initiating
warfarin vs direct-acting oral anticoagulants (DOACs). Methods: We
included AF patients with no history of diabetes newly prescribed OACs
in Clinical Practice Research Datalink (CPRD) Aurum between 01.01.2013
to 31.01.2019. Patients were followed from 30 days after the first OAC
prescription to 21.01.2020, death, deregistration or the first
occurrence of a diagnostic code for T2DM. Potential confounders were
identified using a directed acyclic graph, and Cox regression was used
to calculate adjusted hazard ratios (aHR), comparing the hazard of T2DM
between warfarin and DOAC initiators. We investigated a-priori
interactions with sex and age and conducted subgroup analyses according
to the type of DOAC. An on-treatment analysis was added post-hoc to take
into account treatment switches and discontinuation. Results: A total of
94,394 OAC initiators were included, 46.34% were female, and the
average age was 74.78 (SD 11.6) years. After a mean follow-up of 2.77
years (IQR 1.45-4.19), 7,041 patients developed T2DM, with the crude
Kaplan-Meier risk slightly lower among DOAC initiators (13.58%,
95%CI=12.89 – 14.31) compared to warfarin initiators (15.53%, 95%CI
= 14.99 – 16.09). After adjusting for a-priori identified confounders,
a modest protective association between DOACs and incident T2DM was
found (aHR=0.90, 95%CI 0.84-0.95; p<0.001). No interaction by
age group or sex was observed (p=0.11 and 0.61, respectively). Subgroup
analyses were consistent, with potentially somewhat more marked
protective effects for initiators of edoxaban and dabigatran compared to
warfarin, although confidence intervals were wide [HR=0.82 (95%CI
0.63-1.08) and HR=0.86 (95%CI 0.76-0.96), respectively]. Sensitivity
analyses applying a lagged index date (initiation + 180 days) led to
consistent results. On-treatment analysis showed a more marked
protective effect, with an adjusted HR of 0.69 (0.65, 0.73).
Conclusions: Treatment initiation with DOACs compared to warfarin was
associated with a relatively modest reduction in the risk of developing
T2DM in this cohort. We did not find any evidence that the association
differed according to patient sex or age.