Oct 27, 2012
I've talked quite a bit about atrial fibrillation on ERCast, and the topic that generates the most emails, by a wide margin, is cardioverting the hemodynamically stable patient with recent onset atrial fibrillation or flutter (RAFF). Recent onset is defined by dysrhythmia onset within 48 hours of presentation. Forty eight hours is considered the time window for safe (minimal thromboembolic risk) RAFF cardioversion.
The palpitations, shortness of breath, dizziness, and weakness caused by RAFF were the reason they came to see you. Having done hundreds of ED RAFF cardioversions, I can tell you, that when patients get out of RAFF, they feel much better. Leaving a RAFF patient in a fib presents an extra level of complexity. What is your plan for discharge if you are going to leave the RAFF patient in atrial fibrillation? Most of the time, it involves anticoagulating and referring them to cardiology for cardioversion at a later date. Is that really a better option? If it were a matter of patient safety, i.e. performing delayed cardioversion with full anticoagulation is safer than ED RAFF cardioversion, then there would be no discussion-we would err on the side of patient safety. But there is no evidence that delayed cardioversion with full anticoagulation and echocardiogram to rule out clot in the atrium is any safer than cardioverting RAFF patients in the ED.
There are several options for the hemodynamically stable RAFF patient in the ED.
Option one: Have the patient come back tomorrow so they can be rechecked within 48 hours of symptom onset, since that is the known window of safety for cardioversion without anticoagulation. Fifty percent of RAFF patients will convert spontaneously within 24 hours and there is no rule that says cardioversion has to happen this minute. I instruct patients to not eat after midnight (so they can have an empty stomach for procedural sedation-not that there's evidence to support this, but I'd rather have a patient with an empty stomach than one with a belly full of Egg Mcmuffins and Vegemite), come back in the morning and, if they’re still in a fib, proceed with cardioversion. If there is rapid ventricular response associated with the RAFF, I will IV rate control in the ED and give an oral rate control agent (usually diltiazem) prior to discharge.
Option two: Cardiovert right now- either chemically with procainamide or straight to electricity. I’d say about 80% of the time, the experienced a-fibbers want to go straight to electricity and not mess around. They also don’t want to have to come back for a recheck. The new onset a fib patients will opt for the 'come back the next day ad recheck' option more often than a patient with previous episodes, but even the majority of first timers go for cardioversion during the current ED visit. I like the Ottawa Aggressive Protocol for managing RAFF and will offer 1g IV procainamide over 1 hour as an initial step, especially if the patient is hesitant to proceed with sedation and electricity.
Option 3: Do no ED cardioversion at all. Anticoagulate (or start on a daily aspirin if a low CHADS score and no plan for future cardioversion) and refer the patient to cardiology. You can also start your patient on LMWH and set them up for an urgent TEE to look for atrial clots. Some hospitals do not have ready access to transesophogeal echocardiogram (TEE) and, in this case, the patient will get a few weeks of anticoagulation before cardioversion.
There is an option 4, which doesn't get the bold highlight, and that is to rate control, give a dose of low molecular weight heparin, and admit all RAFF patients to the telemetry unit. This is still done in many hospitals in the United States, but I'm not sure why.
Let’s address the elephant in the room...
How do you really KNOW there is no clot in the atrium? How do you know you aren’t going to stun the atrium, make some sort of thrombogenic medium and cause a stroke in a few days? Is there a way to know this without doing a TEE? A study in Heart 2011 suggested that N-terminal pro-brain natriuretic peptide (NTproBNP) level can help to tell if patients have been in afib for a short or long period of time and whether or not there is a clot in the atrium. This was a small study and may ultimately allow us to do cardioversions in patients who aren’t sure of how long they’ve been in afib, but there’s not enough evidence to support it NTroBNP in clinical practice. Use of D-dimer to detect atrial clot has been tried with varying success but the evidence isn’t convincing. But why even worry about biomarkers? If the patient knows exactly when their symptoms started and they present within 48 hours of onset, do we need to go through any extra steps beyond an H and P and having a conversation?
Let’s take a look at a December 2011 Annals of Emergency Medicine article titled: Is discharge to home after emergency department cardioversion safe for the treatment of recent onset atrial fibrillation? This was a “Best Available Evidence” review that looked 5 papers addressing the safety of ED cardioversion.
The authors's synopsis of the 5 reviewed papers was that most of the complications related to cardioversion come from procedural sedation. When I am having a PARQ (Procedures, Alternatives, Risks and Questions) conversation with the patient, one of the main things I focus on is how we are going to do the sedation, the potential risks and what we will do to mitigate those risks.
As far as complications from cardioversion itself, in the five studies reviewed, there were some chest wall burns and 2 episodes of ventricular tachycardia. One v-tach converted spontaneously and 1 was successfully shocked. Sometimes, in the back of our minds, we worry that we are going to cause a more serious arrhythmia if we shock a patient with a-fib. There have been a few reported cases of ventricular tachycardia but it is extremely rare, and the two in this paper were short lived.
Combining the outcome data of the 5 studies in this 'best available evidence' paper, there were zero reported post cardioversion thromboembolic events after ED discharge with follow up periods ranging from 7 to 30 days. Zero is pretty impressive, and when we’re talking about something like stroke, zero risk is about where we want to be. But like most things in medicine and life, there is no 'all or nothing' and the risk cannot really be zero.
The first study that directly looked at the risk of stroke and conversion to sinus rhythm in the setting of RAFF was from Annals of Internal Medicine 1997 titled Risk for Clinical Thromboembolism Associated with Conversion to Sinus Rhythm in Patients with Atrial Fibrillation Lasting Less Than 48 hours. Three hundred fifty seven hospitalized patients converted to sinus rhythm within 48 hours of a fib onset. Some converted spontaneously and some were cardioverted. Out of those 357, there were three thromboembolic events shortly after conversion from fib to sinus. All three of these patients were in their 80s and all three converted spontaneously.
There are several theories addressing why atrial fibrillation causes clots to form in the atrium. One is that prolonged fibrillation causes stagnation of blood and subsequent clot formation. Another is that conversion from fib to sinus, either spontaneously or via cardioversion, depresses left atrial function which leads to clotting. We worry about both of these when considering cardioversion of ED RAFF patients. Regarding theory two, clot formation as a result of cardioversion, wouldn’t it make sense to give some anticoagulation like LMWH at the time of cardioversion to decrease the risk of clot formation? In the 1997 Annals of Internal Medicine study mentioned above, no difference in thromboembolic stroke risk was seen between patients who did and did not receive acute anticoagulation with either warfarin or heparin. Take that with a grain of salt because this study was not powered to define/detect a small difference in benefit from anticoagulation in RAFF. The 1997 Annals study showed a 0.8% rate of thromboembolism and most others report a 0.0% rate. A clinical trial designed to detect a decrease in thromboembolic risk would need thousands of patients with treatment and non treatment arms to see if there's any benefit giving anticoagulation for RAFF cardioversion. At that point, when you are giving thousands of patients acute anticoagulation to try and prevent a rare event, you’re likely to see more harm from the treatment itself than benefit in preventing strokes. Some, such as the American College of Chest Physicians, recommend considering acute heparin for high risk patients getting cardioversion, but even they recognize that there’s no evidence to support this approach.
When discharging a patient with new onset atrial fibrillation after a successful cardioversion, there are a few key points I put into the conversation.
"Atrial fibrillation is usually not a one time occurrence and will invariably recur. The next time it happens, you don't need to come immediately to the ED if you aren't having chest pain, SOB, feeling dizzy, etc. Give it some time if you can tolerate the symptoms. Most episodes are short lived and, for the longer ones, there is a 48 hour window in which we can cardiovert.”
"You need to establish care with a cardiologist to get an echocardiogram and discuss long term a fib management."
If this is a patient with new onset a fib or one who is not on an anti-platelet agent or anticoagulant, I will also calculate the patient’s CHADS VASC score on MDcalc and give them a printout showing their future stroke risk. Many of these folks are not even taking aspirin - I at least start them on that. Should you start oral anticoagulants in the ED? I personally don’t but have them follow up with cardiology or their primary provider ASAP. The decision of whether or not to start anticoagulation is a huge conversation. One of my partners feels differently. He starts patients on anticoagulation right there in the ED if patients are high risk. But this, my friends, is a talk for another day.
Recommendations mentioned in podcast
Of the 50 ERCast episode's to date, my favorite...
Top 2 ERCast episodes by number of downloads