番茄社区

BOSTON -- Lower adherence to non-vitamin K antagonist oral anticoagulant (NOACs) might wipe out the advantage over warfarin (Coumadin), according to a large real-world observational study.

In an unadjusted model, patients with a filled prescription more than 80% of the time had higher risk of hospitalization or an emergency department visit for stroke or systemic in the 2.5 years after the index prescription if on a NOAC versus warfarin (P<0.001).

After adjustment for prior stroke, hypertension, age, and other key risk factors, that difference reversed and significantly favored a NOAC, as had been the case in the randomized trials, Dhanunjaya Lakkireddy, MBBS, of the Kansas City Heart Rhythm Institute and Research Foundation in Overland Park, Kansas, reported here at the Heart Rhythm Society meeting.

However, lower NOAC adherence showed the highest thromboembolic event rate (4.05 per 100 patient-years) and was consistently associated with higher risk in both adjusted and unadjusted analyses. After controlling for other factors, filling a NOAC prescription 40% to 80% of the time studied had a hazard ratio of 1.298 (95% CI 1.100-1.532) compared with high adherence warfarin.

Low NOAC adherence carried an adjusted relative risk of stroke of any kind similar to that of low warfarin adherence.

Because low adherence to warfarin has been so feared both for bleeding and thromboembolic risk, "it's rather ironic we would postulate the notion that a non-adherent patient should go on warfarin. It's sort of backwards," session co-moderator Andrew Krahn, MD, of the University of British Columbia in Vancouver, said at a press conference for the late-breaking clinical trial.

"There are limitations to these association datasets, administrative datasets, that we don't try to make a lot of therapeutic recommendations on [them]," he cautioned.

Lower overall efficacy of the NOACs would contrast to nearly all other real-world observations, study discussant Hein Heidbuchel, MD, PhD, of Belgium's Antwerp University noted.

Bias by indication could be at play, he cautioned, if the highest risk are put on the newer class believed to have lower risk for stroke. That would only emerge with deeper digging, he said.

Counterintuitively, lower adherence was associated with greater bleeding risk in the cohorts (HR 1.434 for NOAC and 1.323 for warfarin versus high-adherence warfarin, both P<0.05).

The explanation for this appeared to be bias by indication, as a before-versus-after comparison of adherence showed that adherence dropped after a bleeding event in both groups.

The analysis included a strictly defined group of 83,168 non-valvular atrial fibrillation patients with elevated stroke risk as the only reason for oral anticoagulant prescription in 2014 or 2015 in the IBM Watson Health MarketScan administrative claims database.

In it, 36.9% of the warfarin group had a prescription filled 40% to 80% of the time studied, as did 26.5% of the NOAC group. Patients with less than 40% adherence were excluded to minimize heterogeneity and reduce confounding in the data, Lakkireddy said.

Limitations included lack of propensity matching or adjustment for the different dosing strategies between drugs that could affect frequency of prescription fills required for adherence.

Heidbuchel noted that the administrative data could not say whether the prescribed drugs were actually taken correctly, although prescription days covered is a "reasonable proxy for adherence."

At this point, "it's hard to make any sound conclusion" based on the data, Heidbuchel cautioned. "We need some further analysis on this dataset. But what is clear and stands out is the adherence issue, which is something we have to think of, we must address."

Comment