Incidence of bleeding and thrombotic events in non-institutionalized paediatric patients using warfarin in the united kingdom

Abstract

Background: Dosing of vitamin K antagonists (VKA) is complex with large inter- and intra-individual variability in patients' required VKA dose. Over- and underdosing can result in bleeding and thrombotic events. The incidence of these events in paediatric patients on warfarin therapy in a European population is unknown. Objectives: To estimate the incidence of bleeding and thrombotic events in warfarin using paediatric patients in the UK and to characterise patients who do or do not experience a bleeding or thrombotic event. Methods: Data were obtained from the UK CPRD in the period between January 1998 and November 2016. Using a cohort design, we identified all patients with ≥1 prescription for warfarin and who were ≤18 years. The date of the first prescription marked the start of the follow-up. Follow-up was classified into periods of warfarin use and non-use. Patients were followed until 19 years of age, death or departure from the practice. The incidence of non-fatal bleeding and thrombotic events was assessed using both information from CPRD and the linked Hospital Episode Statistics (HES). Fatal events were identified usings the linked mortality data from the Office for National Statistics (ONS). For calculating the incidence of thrombotic events only patients without a history of thrombosis were included. Results: In total, 685 patients were identified (median age 15 years, 45.4% female) of whom 372 could be linked to the HES and ONS databases. The incidence of bleeding and thrombotic events during warfarin use was 4.08 and 1.27/100 patient years, respectively. The incidence of bleeding events during non-use was 2.65/100 patient years (relative risk 1.58, 95% confidence interval [0.89-2.80]). Only 2 fatal events occurred, one bleeding and one thrombotic event. Patients with a bleeding event tended to have a higher percentage of INR measurements with a value above 4 (9.4 vs 3.9%) and a lower fraction below 2 (18.4 vs 39.1%) compared to patients without a bleeding event during the whole follow-up. Patients with a thrombotic event showed the opposite trend, a higher percentage of INRs below 2 (45.8 vs 29.5%) and a lower percentage of INRs above 4 (2.7 vs 5.3%). All differences were not statistically significant which maybe due to the small sample size. Conclusions: The incidence of bleeding events was higher than of thrombotic events. The trends in percentages of INRs under and above therapeutic range suggest that keeping the INR within range could decrease the occurence of these events.