Abstract Warfarin has been associated with bleeding and venous thromboembolism.

Objective:

This study aimed to estimate the association between bleeding and concomitant self-medication, and the incidence of adverse drug reactions in patients using warfarin.

Setting:

the public health network of Ijuí, a municipality in southern Brazil. This was an open prospective cohort, conducted for a period of 18 months with users of warfarin, treated at the public health service. The association between bleeding and self-medication was evaluated by means of the Cox Model with left truncation, using the time variable.

Main outcome measurement:

bleeding reported in the follow-up. Cases of thromboembolism and death were also registered. All patients treated with warfarin in the public health system of the municipality (98) were identified. Sixty-eight were interviewed and followed up, of whom 63 completed follow-up and five died during the study. Bleeding rates of 37.7 /100 patients/year, thromboembolism of 4.8/100 patients / year and deaths of 4.8 /100 patients / year were observed. The results showed that patients, who take warfarin and self-medicated present a two-fold increased bleeding, compared with those who do not self-medicate. The bleeding risk associated with self-medication ranged from 2.001 to 2.685; those values maintained their significance even when adjusted for number of interactions, CYP polymorphism, TTR and age in COX analysis. These results greatly suggest the need for providing greater assistance to patients who take anticoagulant medications with the purpose of reducing self-medication and consequently, adverse reactions.