Optimal adherence thresholds for oral anticoagulants in patients with atrial fibrillation using machine learning and population administrative data

Read the full article See related articles

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

Background

Adherence to oral anticoagulants (OACs) for atrial fibrillation (AF) stroke prevention is traditionally defined as taking 80% of doses as prescribed, but this threshold has not been validated in terms of clinical outcomes.

Objectives

We sought to determine the OAC adherence thresholds maximally associated with risk of clinical events in patients with AF. We also sought to compare these thresholds to the conventional 80% threshold.

Methods

This was a cohort study using retrospective data of patients newly diagnosed with AF who were new users of OACs from 1996 to 2019 from population-based administrative data from British Columbia, Canada. We built Cox proportional hazard models with OAC adherence, measured as proportion of days covered (PDC), as the main exposure captured during 90 days before outcome events or end of follow-up and applied LASSO estimated coefficient paths to identify candidate thresholds, followed by steps to identify which were truly optimal in terms of outcomes.

Results

A total of 44,172 patients were included. For all outcomes, VKA optimal PDC thresholds were between 0.85 and 0.95 and the DOAC optimal threshold was 0.9. Outcome hazard reduction for adherence above vs. below the optimal threshold was greater for DOAC than VKA. For DOAC, PDC ≥0.9 yielded risk reductions of 19 to 52% compared to being below it for stroke outcomes, while for these were not significant for VKA for most outcomes. The conventional PDC 0.8 threshold had inferior model fit and clinical relevance compared to the optimal thresholds identified.

Conclusions

Optimal adherence thresholds are higher than conventionally assumed and, especially with DOACs, are strongly associated with clinical outcomes. Our results highlight the need to consider updating the definition of nonadherence to PDC <0.9 for DOACs, and to PDC<0.9 or <0.95 for VKA, depending on the outcome.

Article activity feed