In this ACCEL interview, Ori Ben-Yehuda, MD, FACC, is interviewed by former JACC Editor-in-Chief Anthony DeMaria, MD, MACC, on the topic of left atrial appendage occlusion (LAAO) in patients not eligible for oral anticoagulation (OAC).

Dr. DeMaria: Please give us the background of your study.

Dr. Ben-Yehuda: There are very few data on patients who may have contraindications for OAC or at least their physicians are worried enough that they don’t place them on OAC for stroke prophylaxis in the setting of atrial fibrillation (AFib). We know the increasing incidence of AFib due to the aging of the population and comorbid conditions is a problem. We wanted to obtain data on the number of such patients and their outcomes. Do they bleed more? Do they have more stroke? What happened if there were not on OAC?

Intuitively I would have thought we knew what happens to people with AFib who have contraindications to OAC, but as you say those data have not been clear.

They haven’t been clear because studies that assessed OAC and their efficacy have excluded such patients. We believed also that their incidence is not very high, but they’re out there. This is where big data comes in, by providing databases that collect information on millions of patients. We partnered with an entity called Marketplace, now owned by IBM, that collects insurance claims data on millions of patients.

"There are very few data on patients who may have contraindications for OAC or at least their physicians are worried enough that they don’t place them on OAC for stroke prophylaxis in the setting of atrial fibrillation." — Ori Ben-Yehuda, MD, FACC

They have 40 million employer-sponsored insurance records, and Medicare supplemental claims data for another 3.4 million. They have information on inpatient and outpatient visits with diagnoses and pharmacy claims, which is very important. We know whether patients were on OAC or whether they filled a prescription.

Tell us about the database and the patient demographics.

We looked at a time period between 2010 and 2012 and included patients who had at least one diagnosis code for AFib that was not within 30 days of CABG. We then looked at possible contraindication events, including the most severe (intracranial hemorrhage, intraocular hemorrhage, gastrointestinal hemorrhage), and conditions that would be contraindications that are rarer but recorded in the database, such as untreated (type B) aortic dissection.

We identified 43,000 patients who had these contraindications in the setting of AFib who were not prescribed either OAC or low-molecular-weight heparin.

What did you observe?

There was a correlation between the underlying CHADS₂ score and CHA₂DS₂-VASc score and risk of stroke, which increased with increasing scores. We also found these patients, not unexpectedly, had a high bleeding risk even though they’re not on OAC – about 20 percent had another bleeding episode.

I think that’s very important because it justifies their physician’s decision not to place them on OAC. Also, we found a very high risk of ischemic stroke during follow-up in patients with a history of intracranial hemorrhage – independent of the CHADS₂ score, but this was limited by the smaller sample size.

In the 20 percent of patients not on OAC who had bleeding, if they had been on an OAC and had bleeding, would it have been attributed to the OAC? Is it possible If everyone had been on OAC that only about 20 percent would have had bleeds and perhaps they’re the same 20 percent that had bleeding off OAC? In other words, is that finding interesting in terms of whether this group of patients should be given a trial of OAC?

It’s a very interesting question. Intuitively, in general, you would think that if a patient has bleeding and then goes on an OAC that the bleeding would be more severe. I think it would be very hard to study this in a controlled manner, as an institutional review board would have a hard time signing on to giving patients who are already at risk a medicine that would increase their risk.

"These data support the notion that there is a group of patients in whom the risk is sufficient to warrant consideration of these devices." — Ori Ben-Yehuda, MD, FACC

In a large database, we can at least match patients who have similar conditions to those in our study on OAC to see the outcomes. However, beyond the risk scores, physicians have an additional intuitive ability. As we try to develop frailty scores, for example, we see these are not as good as the physician’s assessment. You look at the overall patient and you know whether or not they’re frail.

I think that physicians know their patients, and there’ll still be a selection bias, even if we match them on different characteristics.

What did you conclude from your observations? Bleeding was common and ischemic events were fairly prevalent too.

There is an overall a correlation between the CHADS₂ and CHA₂DS₂-VASc scores for both ischemic events, for which they were designed to detect risk, and bleeding events. This puts physicians and patients in a quandary. The more you need these medications, the more you’re at risk for side effects (bleeding) of these medications.

We definitely showed that [patients who are ineligible and therefore not treated with OAC] is not a rare phenomenon. They tend to have relatively high CHADS₂ scores. The vast majority of our patients had a CHADS₂ score >1 and a CHA₂DS₂-VASc score >4. Thus, this is an important group of patients for whom we need to find a solution.

This sets the stage for innovative new therapies that are now approved, such as LAAO to eliminate not the entire risk but at least reduce it to a significant degree and help protect these patients.

These patients were evaluated prior to the availability of occlusion devices. How many would qualify today?

This is a high-risk group and would qualify. The studies for LAAO and other databases had wider inclusion criteria. An ongoing debate is who really needs these devices. Their cost and availability is an issue, but I think these data support the notion that there is a group of patients in whom the risk is sufficient to warrant consideration of these devices.

Is there a CHADS₂ score that helps to define who should have an occlusion device?

A CHADS₂ score of ≥2 identified patients who were at significant risk. As with our decision to implement OAC, this study suggests that in the absence of the ability to give OAC, the patient should be considered for a closure device.

This interview was edited for print from an interview transcript.

Keywords: ACC Publications, Cardiology Magazine, Atrial Appendage, Atrial Fibrillation, Ethics Committees, Research, Follow-Up Studies, Gastrointestinal Hemorrhage, Incidence, Intracranial Hemorrhages, Medicare, Outpatients, Selection Bias, Stroke