Introduction The occurrence of stroke secondary to the pro-thrombotic effects of AF is well recognised, with warfarinisation the established therapy for prevention. This is not without complications and so the introduction of the novel oral anticoagulants (NOAC), Dabigatran a factor IIa inhibitor and Rivaroxaban a Xa inhibitor have been received with enthusiasm. Advantages include single dosing, fewer drug interactions and freedom from monitoring, whilst data has demonstrated non-inferiority with respects to stroke prevention. A reported greater propensity to gastrointestinal (GI) bleeding is concerning given the lack of accepted reversal agents, in a cohort of patients who tend to be elderly with comorbidities. We aim to establish the GI complications associated with NOACs in stroke prevention.

Methods We retrospectively analysed 170 consecutive patients from our TIA clinic, who were found to have AF. They were commenced on either Dabigatran or Rivaroxaban, with a total of 86 patient-years of exposure at the time of analysis. We interrogated our computer-based database to ascertain the incidence of anticoagulation related admissions and subsequent management

Results Our cohort of 170 patients had a mean age of 76 years, with a male to female ratio of 8:9. Dabigatran at 220–300 mg/day was prescribed to 45% (n = 77), whilst 55% (n = 93) received Rivaroxaban 20–15 mg/day. Rivaroxaban was stopped in one patient due to drug induced hepatitis. One patient was switched from Dabigatran to Rivaroxaban due to dyspepsia.

GI bleeds occurred in 5% (n = 4) of the Dabigatran group, with 2 upper and 2 lower GI bleeds, occurring on average after 25 days. These patients had a mean age of 84 years, with at least 2 comorbidities. Endoscopy revealed an oesophageal ulcer requiring clipping, new diagnosis of colorectal cancer, haemorrhoids and in one patient no cause was found. GI bleeding occurred in 1% (n = 1) of the Rivaroxaban group. The patient had a rectal bleed, was 96 years and too frail for investigation, therapy was stopped. The average blood transfusion requirement was 4 units. One individual required resuscitation with 16 units of blood, 12 of fresh frozen plasma, 4 of cryoprecipitate and 1 pool of platelets. Average length of admission was 8.4 days.

Conclusion Existing data suggests that the GI bleed rate whilst on warfarin ranges from 1.3–2.6%, compared to a rate of 2.3–3.6% in association with NOACs. In our series there was an overall GI bleed rate of 3%, with 1.8% that could be considered to be significant, requiring blood products and cessation of anticoagulation. Should NOACs become more widely used the lack of established reversal techniques will require local GI units to provide a prompt and skilled bleed service.

Disclosure of Interest None Declared.