Aug 20, 2021 This Week in Cardiology Podcast

John M. Mandrola, MD


August 20, 2021

Please note that the text below is not a full transcript and has not been copyedited. For more insight and commentary on these stories, subscribe to the This Week in Cardiology podcast.

In This Week’s Podcast

For the week ending August 20, 2021, John Mandrola, MD comments on the following news and features stories.


The COVID situation continues to be focally bad. Hospitalizations in some areas have risen to the point where many places are postponing elective procedures. We are close to that but not quite there. But there are also some mildly optimistic signs: states first hit seem to have seen cases peak. Look at the curves for Missouri, Louisiana, and even Texas. The 7-day moving averages have flattened.

While the world would have been better off without this virus, I want to highlight three positive aspects of the current situation.

  • One positive is the pressure to do same-day discharge after procedures has forced many of us, me included, to break out of our dogma that overnight hospital stays were necessary. Jason Andrade and colleagues from University of British Columbia has published on the feasibility of same day discharge of atrial fibrillation (AF) ablation. The COVID situation has sped translation of this work.

  • Another positive: vaccine efficacy has held up nicely. While we have all heard of rare breakthrough cases, the vast majority of severe COVID cases have occurred in the unvaccinated. Given the near certainty that SARS CoV-2 will become endemic, the main outcome measure now is no longer cases, but hospitalizations.

  • Third positive: the increase in cases has taught two important statistical lessons, and any time statistical lessons reach the public, the world becomes a better place. That’s because numerical literacy drives a lot of terrible decisions.

  • The first lesson is the base rate phenomenon. We’ve learned that in states or countries that have high vaccination rates, most of the new cases of COVID occur in vaccinated individuals. Of course, they do. Even with excellent vaccine efficacy, there are much higher absolute numbers of vaccinated people than unvaccinated.

  • The second lesson is Simpson’s Paradox, where it is possible to draw two opposite conclusions from the same data—depending on how you divide things up. Or another way of saying it: a trend or result is present when the data is put into groups that reverses when the data is combined.

Simpson’s Paradox

A paper from Israel got a lot of attention because it purported to show reduced vaccine efficacy against severe COVID. I will link to an excellent explanation From University of Pennsylvania Professor Jeffrey Morris.

The absolute number of severe cases was 214 per 100,000 among the unvaccinated and 301 per 100,000 in the fully vaccinated. Almost the same; that looked bad. But there are two lurking variables: One is the base rate of unvaccinated vs vaccinated. There were 1.3 million unvaccinated but 5.6 million vaccinated. So now if you adjust for base rate, it’s 16.4 severe cases per 100,000 in the unvaccinated vs 5.3 severe cases per 100,000 in the vaccinated. So, at 67%, much better vaccine efficacy when you adjust by the base rate.

But wait, you have heard vaccine efficacy with the Pfizer BioNTech is > 90% for severe cases. What gives with only 67%? Perhaps it’s the Delta variant. No. There are more lurking variables. Recall that the vast majority of older people are vaccinated compared with 73% of younger folks. In other words, more than 85% of the unvaccinated group are younger—defined as under 50 years old. But also remember the age-gradient of COVID. Older people are much more likely to get severe COVID. That’s another lurking variable.

Now if you break out the vaccine efficacy by age groups you can compute the vaccine efficacy for severe disease for each age group. In younger folks the number of severe cases were 13 times more likely in the unvaccinated; adjusted vaccine efficacy is 92%. In older folks, the number of severe cases were 7 times more likely in the unvaccinated; adjusted vaccine efficacy is 85%.

So, indeed, the vaccine holds up very well against Delta. The paradox here occurs because of the fact that both vaccination status and risk of severe disease are systematically higher in the older age group .

To be sure, there may be other variables, such as early vs late vaccination, but my main point here is not so much vaccine efficacy, but that observational data is super-complicated. While randomization tends to balance both known and unknown variables, observational data requires thinking carefully about all the potential lurking variables. Whenever I see an observational paper making big conclusions I pause and think, well, this is going to require going slow and looking hard at the methods. Understanding things like base rate and Simpson’s paradox inoculates one against being bamboozled by dubious scientific conclusions.

Left Atrial Appendage Occlusion: Two Stories This Week

First, the US Food and Drug Administration (FDA) has given approval for the Amulet device. This is a percutaneous appendage occlusion device that is a bit different in design than Watchman. Same idea though—mechanical stroke prevention. Abbot makes the device and it’s been used in Europe for many years. It is quite popular there.

What’s weird about this approval is that it was based on the Amulet IDE trial, which is to be presented next weekend at the European Society of Cardiology (ESC). FDA surely has seen the data before we have. The Amulet IDE study compares Amulet with Watchman. It had three primary endpoints: a composite of safety endpoints; ischemic stroke and systemic embolism; and device closure.

Each of the three endpoints were tested with a non-inferiority design. I think we will have to see the data, but I am concerned that noninferiority was used in all these endpoints. Would we not want a device that was better in some way?

I would have loved to seen a third arm in this trial—a non-procedural arm: Watchman, Amulut, conventional therapy. That would have taught us so much. The medical arm would have had zero procedural complications.

JACC-Interventions published the second Watchman study, by Keerat Ahuja, MD, and co-authors. They used data from the Nationwide Readmissions database to study the association between chronic kidney disease (CKD) and end-stage renal disease (ESRD) and in-hospital and short-term outcomes in patients who had left atrial appendage closure (LAAC).

Picture three groups in this retrospective observational study of ≈ 21,000 patients who had left LAAC: about 16,000 patients had no kidney disease, about 4000 had CKD, and 570 had ESRD. The findings were not surprising:

  • Patients with ESRD were 6.5 times more likely to die in the hospital after LAAC than those without ESRD. The confidence interval ranged from 3.3 times to 12 times higher risk of death. The absolute in-hospital death rate in patients who had LAAC was 3.3%.

  • CKD was associated with increased risk for in-hospital acute kidney injury or hemodialysis, and stroke or transient ischemic attack.

  • ESRD and CKD were both associated with increased readmissions extending to 90 days compared with no CKD.

While there are wide confidence intervals, and the definition of CKD is fluid, the general trends highlight the added risk of harm incurred by patients with CKD.

This is a good use of registry data. It reinforces that the net benefit calculus of this preventive procedure differs greatly in patients with CKD. A patient considering LAAC must believe that the added benefits of having the appendage plugged and maybe not being on oral anticoagulation (AC) must be greater than a 3% to 9% risk of a major adverse event.

But if you have advanced kidney disease, what this study shows is that the baseline risk is 3 to 12 times greater. That is huge. And consider that patients with CKD have many other competing causes of cardiovascular events.

But perhaps the biggest problem with recommending this procedure to patients with advanced CKD is that we don’t have any idea—literally no idea—whether AC is a net benefit in the first place. This is because so few patients with advanced kidney disease were randomized in pivotal AC randomized trials. Observational studies go both ways in patients with kidney disease. Some observe fewer strokes with AC, others observe more bleeding. But all observational studies are heavily confounded. Even if I were a LAAC proponent, I would not rec it for patients with advanced kidney disease. There is an analogous story here with CKD and implantable cardioverter-defibrillators (ICDs). There is no trial-level evidence showing that preventive ICDs improve mortality in patients with CKD. First, patients with severe CKD were not included in the clinical trials. Second, when you look at patients who had any CKD from the RCTs, there was no ICD benefit. Third, the ICD2 trial of preventive ICDs vs medical therapy in patients with ESRD was stopped for futility.

The obvious reason preventive therapies struggle in patients with severe multi-morbidity is that there is a higher upfront risk and tons more competing causes of bad outcomes.

ESC Preview: EMPEROR-Preserved and the LOOP Study

I will have a preview column up next week before the ESC. Here are some of the studies that I am looking forward to.

The headliner for this year’s virtual meeting will be the EMPEROR-Preserved trial, which compared the sodium-glucose cotransporter 2 (SGLT2) inhibitor empagliflozin with placebo in approximately 6000 patients with heart failure and a preserved ejection fraction (HFpEF). This trial may be the biggest of 2021.

  • We already know from a press release that the trial met its primary endpoint of cardiovascular (CV) death or hospital admission for heart failure.

  • The question to be answered from the presented data is whether statistical significance (which can be inferred from the press release) translates to value and clinical significance.

Clinical significance will turn on critical appraisal of the trial. How robust was the reduction in outcomes? We should always shun arbitrary P-value cutoffs but a P-value of 0.001 is more robust than 0.04.

More important than the P-value is the effect size of the benefit. I want to see the absolute risk reduction and whether the primary endpoint was driven by cardiovascular death or the softer endpoint of heart failure hospitalization? Also, do the subgroups give us a clue to which type of patient may benefit more, or less?

When considering value, I propose two extreme examples: a low-value case would be if the drug barely reaches statistical significance, the primary endpoint is driven by modest absolute reductions in heart failure hospitalizations not CV death, and no overall mortality benefit. A high-value case for empagliflozin would require a robust statistical result, reductions in both CV death and heart failure hospitalizations, as well as lower overall mortality.

The Loop Study. Thus far, digital devices have delivered far more information than knowledge. We can now easily detect minutes to hours of asymptomatic atrial fibrillation (AF). But then what? I get this question often: “Mandrola, my 70-year-old patient has 1 hour of AF; should I start an anticoagulant? And for how long? For life...for 1 hour of AF? Really?

The LOOP study is randomizing about 6000 Danish patients older than 70 years who have one stroke risk factor to AF screening with an implantable cardiac monitor or usual care. The randomization is 3:1; 4500 patients in the ICM arm and 1500 in the control arm. I will have to ask a statistical colleague to comment on that.

This trial excites me because the primary outcome is time to first stroke or systemic embolism. This differs from the many previous AF screening trials that consistently show AF screening (of any type) leads to more detection of AF and more use of anticoagulation.

The question we want answered is, what is the net benefit of treating these episodes? Starting anticoagulants for short-duration AF is considered a good thing. But we don’t actually know if it will deliver more benefit (stroke reduction) than harm (bleeding).

  • The first step in showing net benefit is: does all the added anticoagulation from AF screening significantly reduce stroke? The active arm of the LOOP protocol calls for anticoagulation for any detected AF more than 6 minutes. That is a very low bar. Lots more patients in the active arm will be on anticoagulation.

  • If LOOP shows a reduction in stroke, the next question will be, at what costs? In the trial protocol, major bleeding is listed as an “other outcome,” but I see it as a key outcome.

I will have more studies in my preview column. Steve Stiles will have an even more complete preview of the hot topics and late-breakers.

By the way, the anticoagulation in COVID-19 trial is one of the hardest I have seen in years. Stay tuned.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: