Background:

The aim of the study was to compare if there were any differences between discontinuation of vitamin K antagonists and direct oral anticoagulants and evaluate the factors associated with such discontinuation in newly diagnosed nonvalvular atrial fibrillation.

Methods:

We conducted a prospective cohort study. Patients were followed-up for 12 months. Since the assignment of anticoagulation treatment was not randomized, propensity score weighting was used considering the baseline characteristics potentially associated with the exposure and result. Factors associated with the discontinuation of anticoagulant treatment were analyzed with a weighted Cox proportional hazards model.

A total of 379 patients were included; mean age was 78 years (SD ± 9) and 58% were women. Median follow-up was 362 days (interquartile range: 347-370 days) and 1% was lost to follow-up.

Results:

The model of time to discontinuation based on inverse probability treatment weighting showed a crude HR of 1.40 (95% CI, 0.79-2.48, and of 1.26 (95% CI, 0.75-2,12, after adjustment for age, type of atrial fibrillation, radiofrequency catheter ablation, bleeding, number of chronic medications and number of medical visits during follow-up for the group treated with direct oral anticoagulants compared with the vitamin K antagonists.

Conclusions:

In our setting, anticoagulant discontinuation in nonvalvular atrial fibrillation would not be associated with the type of drug used, age or type of atrial fibrillation. Radiofrequency catheter ablation, bleeding events and the number of medical visits were associated with treatment discontinuation.