AF ablation Atrial fibrillation

Musings on the CABANA trial — AF ablation vs Drugs

Last week at the Heart Rhythm Society meeting in Boston, Dr. Douglas Packer from the Mayo Clinic presented results of The Catheter Ablation versus Antiarrhythmic Drug Therapy for Atrial Fibrillation (CABANA) trial.

I cannot remember a more consequential HRS (Heart Rhythm Society) meeting. Before and after the presentation, CABANA talk dominated conversations amongst colleagues.

I wrote a column directed at an audience of doctors for the | Medscape Cardiology. You can find it here>  CABANA: Initial Thoughts on AF Ablation vs Drugs

CABANA is important. AF ablation is big business. Hospitals, doctors and the companies that make equipment for the procedure make lots of money from AF ablation. What’s weird about all this growth is that we have not had solid evidence that the procedure improves outcomes, such as stroke or death.

That’s the question CABANA addressed.

You take patients with AF who feel poorly and you put half the group on anti-arrhythmic drugs and the other half get ablation. Then you follow them for years and measure how many in each group have strokes, bleeding, death or cardiac arrest. Those four outcomes were called the primary endpoint. CABANA had many other secondary endpoints–how much AF occurred, quality of life, hospital admissions, etc.

CABANA works like a typical clinical trial: ask a question, then randomly assign patients to different treatments and decide what outcome to measure. The random assignment to treatment group is key because in ideal conditions the only difference between the group is the treatment–drugs or ablation.

The authors hypothesis held that ablation would be superior to drugs for reducing the primary outcome. Remember the primary outcome was disabling stroke, death, bleeding or cardiac arrest. Since the groups are randomized, and all other things equal, then if the ablation arm had fewer events, we could say ablation is superior. Or ablation caused the fewer events.

That is NOT what happened.

The number of primary endpoint events in the ablation and medication were very close–not even close to a statistical difference. The first line of the authors conclusion was that “ablation did not produce a significant reduction in the primary endpoint and all-cause mortality.”


After Dr. Packer presented the main results he then showed results depending on treatment received. In procedure trials some patients cross arms. (If drugs did not work, patients could have ablation, for instance.) So being randomized to an arm does not mean you have to stay in that arm. About 27% who were in the drug arm crossed over to the ablation arm while 9% in the ablation never received the ablation (mostly because of a change in mind.)

The authors knew this would happen and planned for it statistically. But crossovers do complicate the analysis.

Dr Packer showed the as-treated event rates and here the results favored ablation–with lower rates of the primary endpoint. This led some leaders in the field to say ablation was indeed superior.

But you can’t say this.

Statisticians call the use of as-treated comparisons scientific malfeasance. This sounds harsh, but the reason they say it is because when you analyze by treatment received you don’t preserve randomization. And, when groups aren’t randomized, you don’t know whether it was the ablation that caused the improvement or some other confounding factor.

The as-treated analysis essentially becomes an observational study — which make only associations, but not causation claims. (For instance, there are numerous observational studies showing that patients have fewer strokes after ablation, but no one uses these as proof because in observational studies, the reason is likely that healthier people have catheter ablation.) In CABANA, the patients that crossed over were likely different from those that stayed in the drug arm.

A Cautious Interpretation of CABANA: 

One way to interpret CABANA and preserve scientific rules is to say the trial did not actually compare ablation vs drugs as was intended but rather it compared the strategy of ablation vs initial medical therapy with ablation for recurring symptoms.In this scenario, there can be no claim that ablation is superior to drugs for reducing major outcomes, but we can preserve the possibility that ablation remains a reasonable option for selected patients with AF.


Packer presented the safety outcomes of ablation vs drugs. He did not make any formal comparisons, but I tallied the total number of each and there were far more complications in the ablation arm. Surprisingly, there were no deaths related to the procedure, no atrial-esophageal fistulas and no major strokes. This is surprising because about a 1000 people had AF ablation.

Two things to remember about safety is that CABANA was done in experienced centers. I doubt that real-world everyday AF ablation is as safe as it was in CABANA. Second, the complications of AF ablation can have immense downside; they are asymmetric. When a middle-aged person has a stroke, or death from the procedure, this is especially tragic because he or she would surely not have died anytime soon of the AF.

The last thing to say about these results is that they are not yet published in a journal. That means it has not made it through peer review. We will learn more from the published paper.

Where to go from here: 

First, since CABANA found no significance difference in outcomes, we are left with the idea that ablation is done to improve symptoms or quality of life. This is a good goal for any treatment, but it’s hard to sort out how much ablation helps when patients and doctors know who received it. It’s why drug trials have a placebo control.

At least three investigators I spoke with are now planning a placebo-controlled sham trial. In this sort of trial, patients and caregivers are blinded to whether or not an AF ablation is done. Half the group will get a basic EP study, and ablation of arrhythmias on the right side, and no ablation in the left atrium–where AF comes from. The other half gets regular AF ablation. Then no one knows, and the patient is followed.

The history of such placebo controlled procedures have shown that many interventions we thought worked did not do any better than a placebo procedure–renal denervation, vertebroplasty, and placing stents in coronaries for relief of angina.

I know, this may upset some of you. Indeed, a quick-thinking response to doing a sham-control trial is that is unethical to expose patients to a fake procedure. But when you think harder and slower, you realize that the far greater ethical dilemma is exposing millions of people to a risky procedure that produces mostly a placebo effect.

Second, CABANA does not change my current clinical practice. I see no reason to offer more or less AF ablation. I will still do the same general things:

  1. Assess the patients to find out why AF is occurring. Then treat that problem. If it’s obesity or sleep apnea, AF gets better with weight loss and better sleep. If the AF occurred because of transient stress, wait for the stress to past. If the AF occurs because of alcohol, counsel the person to drink less. If AF occurs because of excess training, reduce training.
  2. Remove fear and educate patients. This is the biggest thing. Many patients want stuff done because they fear AF. We teach them about AF and we warn that often times the most dangerous thing about AF can be our treatments. This sometimes takes multiple visits.
  3. If the patient has risk factors for stroke (age, high blood pressure, diabetes, prior stroke, heart failure, blockages elsewhere), we counsel on the benefit/risks of anticoagulation.
  4. If the AF is causing an excessive pulse, we give drugs to slow the rate, so as to prevent heart failure.
  5. If AF persists and if it causes symptoms, then we try drugs. I often use drugs in as needed way. For people with AF once or month or so, they can take an AF drug at that time.
  6. I reserve AF ablation for patients who continue to have AF despite treatment of risk factors and use of AF drugs. I remain a cautious AF ablation doctor.

CABANA has not changed this approach.

Why is Science Emotional?

The final thing to say about CABANA is that its failure to show benefit in outcomes caused some to be disappointed, sad even. Some acted as if it was a failure.

This is a weird view of science. CABANA and trials like it deserve our respect whether or not they show benefit of an anointed procedure or drug. Scientific experiments should be designed to find truth. If the truth is that AF ablation is not better than drugs, then that is really good to know.

Our goal as doctors is not to instrument as many people as possible, but to help as many people as possible.

Yes, there are some purists who say the trial was not powered to find benefit–meaning, the null results of CABANA could be wrong and there was actually a benefit, but the trial did not detect it because it did not have enough patients or had too many people not getting the treatment they were assigned.

This is a reasonable argument, though one must remember that if a treatment is really superior, it should not take thousands of patients to show that it works. Saying a trial found no difference  because it is underpowered often means that the two treatment arms are minimally different.


PS: I will not accept comments that contain personal anecdotes of AF. These are problematic because one of the first rules I tell patients is not to compare their AF or their AF treatment to others. AF affects people in many different ways.

5 replies on “Musings on the CABANA trial — AF ablation vs Drugs”

Thanks for this great patient-centric presentation on CABANA and a reasonable AF ablation strategy. I will be pointing my patients and readers to this URL when I discuss ablation as a possible option.
I have come to believe that AF patients would greatly benefit from having a nonproceduralist AF specialist taking care of them. This would take the “man (or woman) with the hammer” and economic gain issues out of consideration.

Thank you for your insight and wisdom. Very well said. As a non EP cardiologist your approach to afib patients is similar to mine. Thank you for reinforcing that approach.

Great update! Agree, the problem of blinding WRT outcomes assessments and the power of the sham treatment are well established, especially for outcomes that only the non-blinded participants can report. However, isn’t one big symptom/quality of life issue related to a-fib quantifiable in a relatively straightforward objective way- ie how frequent are episodes? Perhaps not all patients find that decreased frequency of a-fib means an improvement in quality of life, but I know that some certainly do. Would it not be possible to use blinded observers to assess the outcome of “frequency of atrial fibrillation” as detected by event recorders or similar method? Annoying for patients to have to wear event recorders for extended periods, but not insurmountable. I have some background in clinical research, but not cardiology, so excuse any ignorance. Perhaps this has been examined already?

Comments are closed.