John M. Mandrola, MD
DISCLOSURES | February 04, 2025
Add to Email Alerts
One of the most interesting trends in modern cardiology is the reversal of dogmas. Routine beta-blockers after myocardial infarction may no longer be needed despite being established practice for years.
The net benefit of oral anticoagulation in patients with atrial fibrillation (AF) may be another area of reversal. Central to this may be new knowledge about untreated stroke risk in patients with AF.
Declining Risk for Stroke in Untreated AF
There is growing evidence that the risk for stroke in patients with untreated AF is declining. And that the risk is a lot lower than predicted by the widely accepted CHA2DS2–VASc score. Recall that the use of anticoagulation (with warfarin) in patients with AF was established in trials conducted more than 30 years ago.
A notable revelation from recent trials of device-detected AF — LOOP, NOAH-AFNET 6, and ARTESIA— has been the low rates of stroke in the placebo arm. Despite mean CHA2DS2VASc scores of 4 in enrolled patients, stroke rates in the placebo arms of these trials were approximately 1% per year. That’s 75% less than what the CHA2DS2VASc score predicts.
One explanation for this discrepancy is that short-duration AF (1-3 hours) detected on a monitor is a different entity from symptomatic AF confirmed on a standard ECG. Both involve fibrillatory activity of the atrium but the former (subclinical AF) clearly confers a lower stroke risk than the latter (clinical AF).
Perhaps the lower risk is simply due to its short duration, but it also could be that some amount of subclinical AF is part of normal aging and has always been present, and does not confer an increased risk of stroke.
But what if the unexpectedly low rate of stroke is due to a more systemic reason — a temporal decline in stroke risk in the population? In patients with heart failure due to reduced ejection fraction, sudden arrhythmic death has declined over time as medical therapy has improved. Why couldn’t stroke risk in AF follow a similar pattern given better management of risk factors, improved air quality, et cetera?
Finnish Observational Registry Study
A recent study using data from the Finnish national registry suggests exactly that.
The investigators identified about 130,000 patients with new-onset AF dating back to 2007. To study the rate of stroke in patients with untreated AF, the authors split patients into three groups according to year of diagnosis and ended up with about 40,000 patients in blocks of 3-year intervals (2007-2010, 2011-2014, and 2015-2018). National registries are useful because stroke events can be correlated with prescriptions filled.
There were three main findings:
- The age and risk profile of those diagnosed with AF increased over time. The mean age went from 70 to 73 years, and the CHA2DS2VASc score from 3 to 3.5 comparing the earliest group to the latest one.
- Despite the worsening risk profile, the overall rate of ischemic stroke decreased by 25%. Absolute rates went from 36.7 to 27.6 events per 1000 patient-years. The decrease was mostly driven by a 32% decline in stroke rates in women vs a 7% reduction in men.
- Most of the decrease occurred in older women with higher stroke risk scores. The rate of ischemic stroke in patients with a CHA2DS2-VA score of 1 (excluding sex) remained stable at approximately 8.2 per 1000 patient-years.
Implications for Interventions and Quality Metrics
Although the study was carefully done, it was observational and retrospective. We should be cautious in its interpretation.
Yet its main strength is that these real-world data comport with those of recent trials, which also show unexpectedly low stroke rates.
These observations don’t tell us who or who not to treat with oral anticoagulation. Those are individual decisions that require judgement and alignment with a patient’s values. The Finnish observations suggest that one reason the placebo group stroke rates in recent trials were far lower than predicted by risk scores is because the population stroke risk is lower.
Taken together, this evidence should reinvigorate our skepticism about knowing stroke rates in untreated AF. I write “reinvigorate” because we should have always been humble about stroke prediction in patients with AF.
We widely accept guideline recommendations and quality measures regarding risk-based oral anticoagulation. In doing so, CHA2DS2-VASc has morphed into something akin to mathematical truth. Little promoted, however, was that the seminal paper establishing the score included just 1000 patients and 25 stroke events.
Also not widely cited is an important meta-analysis of 34 studies involving patients with untreated AF. This remarkable analysis revealed wide variations in stroke rates depending on the cohort. Annual rates ranged from as low as 0.45% to as high as 9%. Most often, stroke rates were lower than established thresholds for anticoagulation benefit.
My point is not to argue against oral anticoagulation for stroke prevention. Rather, it is to infuse us with humility about stroke prediction in patients with AF. Things have clearly changed since the time of the original warfarin studies.
I can however make three conclusions from the recent data.
First, quality measures with strict thresholds for anticoagulation based on integer scores look even more silly. Guidelines need an immediate update. The writers could say something like, “Sorry — recent data obscure the threshold to recommend oral anticoagulation.” And we need another large, placebo-controlled, randomized trial of oral anticoagulation in AF — perhaps with longer-duration AF episodes as an entry criteria.
Another area this change affects is percutaneous left atrial appendage closure (LAAC). Proponents of the procedure often offer it to patients who struggle with oral anticoagulation claiming that the risk for stroke is too high to just stop the drug and do nothing. Well, maybe the risk for stroke is not as high as once thought. Consider, for instance, a trial of LAAC vs no treatment in older patients with AF and co-morbidity.
Finally, the most important conclusion to make is that the evidence that guides practice should have an expiration date. In addition to studying new therapies, we should also study old therapies and dogmas. As technology advances, base rates of bad outcomes of disease change. This surely affects net benefit. We need to have the humility to accept equipoise and do new trials.
John Mandrola practices cardiac electrophysiology in Louisville, Kentucky, and is a writer and podcaster for Medscape. He espouses a conservative approach to medical practice. He participates in clinical research and writes often about the state of medical evidence.
https://meet.google.com/call?authuser=0&hl=en-GB&mc=KAIwAZoBFDoScGludG9fd3RydnBpa3cwa3ZyogE7GgIQADICUAA6AhABSgQIARABWgIIAGoCCAFyAggBegIIAogBAJIBAhABmgEEGAEgAKIBAhAA4gECCACyAQcYAyAAKgEwwgECIAHYAQE&origin=https%3A%2F%2Fmail.google.com&iilm=1740670630233
Leave a Reply