# Prediction of long-term adherence to direct oral anti-coagulants in patients with atrial fibrillation using first-order Markov models

## TL;DR

This paper introduces a new method using first-order Markov models to predict long-term adherence to blood thinners in patients with atrial fibrillation.

## Contribution

The novel use of first-order Markov models to predict adherence to DOACs over years 2–5 based on first-year data.

## Key findings

- Missing even 1 day of treatment per month in the first year predicts lower adherence in years 2–5.
- Adherence increases with age but plateaus around age 75.
- The model demonstrated good calibration for predicting adherence deciles.

## Abstract

Direct Oral Anti-Coagulants (DOACs) are the primary treatment for the long-term prevention of stroke in patients with atrial fibrillation. Strict adherence to DOAC therapy is crucial and must be maintained over the long term. Therefore, predicting long-term adherence is valuable for identifying patients at risk of non-adherence. We developed a novel method for predicting long-term adherence using first-order Markov models to assess adherence in new DOAC users during years 2–5. The prediction utilized age, CHA2DS2-VASc score, and first-year adherence data as predictors. Adherence was measured by calculating the proportion of days covered within consecutive 90-day windows, which were then stratified into deciles. We subsequently calculated the probability of a patient being in a specific adherence decile. The developed model demonstrated good calibration. We discovered that missing even 1 day of treatment per month in the first year was predictive of a lower likelihood of achieving the highest adherence decile in years 2–5. Additionally, we noted a non-linear relationship between age and adherence; adherence increased linearly with age but plateaued around age 75. This innovative approach to modelling and predicting adherence to DOACs for long-term therapy can help identify patients at risk of low adherence and may be applicable to other chronic medications.