Jul 23, 2021 This Week in Cardiology Podcast

John M. Mandrola, MD


July 23, 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 July 23, 2021, John Mandrola, MD comments on the following news and features stories.

COVID-19 Messaging

Cases are rising in the United States and the European Union. Areas with low vaccination rates are doing worse. The New England Journal of Medicine published a big study from British authors showing that the Pfizer and AstraZeneca vaccines have close to similar efficacy against the Delta variant compared with the Alpha variant. It remains to be seen whether this uptick will cause the same percentage of hospitalizations that any previous rise in cases caused. I hope the millions of vaccinated folks will provide a ceiling for hospitalizations.

The obvious challenge now is how to increase the percent of vulnerable adults who are vaccinated. This is why trust in public health officials and their messaging was so critical.

Harvard epidemiologist Michael Mina had some nice comments in New York Magazine. In an interview about the NY Yankees’ positive test story, Mina bemoaned black and white binary messaging. He was talking about the limitations of super-sensitive PCR test and called binary messaging immensely destructive.

One of Mina’s core beliefs in public health is that you absolutely need to bring the public along. You need to keep them informed. If you don’t have the public buy-in for everything you’re doing, you will never defeat a pandemic.

Throughout this pandemic, we’ve generally considered the public to be the problem. But this is public health. The public isn’t the problem – that’s on the virus – instead, the public is the solution.

Pause on that: the public is the solution. assuming that the public was unable to deal with nuance, we have done immeasurable damage.

Doctors who care for patients with chronic conditions know this concept well. The first step is understanding the patient and his or her preferences; then you have to meet patients where they are; and finally, you must have frank discussions of uncertainty and risks.

I see a perfect analogy between how a doctor uses shared understanding with a patient and how a public health expert uses shared understanding with the public. One major lesson of the pandemic will be how to improve public health messaging so as to make the public part of the solution. I agree with Dr. Mina. As it is in the clinic, nuance is crucial, as is humility, absolute frankness about uncertainty, and avoiding any foolish recommendations. (For example, putting 3-year-olds in day care in masks.) Trust is hard won but easily lost.


JAMA-Internal Medicine has published another coffee study. I’ve written previously that the research enterprise would be better off with fewer observational studies of coffee, chocolate, blueberries, etc. For one, these studies are methodologically challenged because of recall bias (people have to self-report); for another, coffee drinking and dark chocolate consumption track with other behaviors, and it’s hard to sort causality out of a single exposure. Plus, I believe the story on coffee is well established. Oodles of recent studies show no association with arrhythmia.

But I learned from the intro section of the JAMA-IM paper that some professional guidelines still suggest avoiding caffeine to diminish the risk of arrhythmia. Who knew guidelines were so slow moving? This particular study from the UC San Francisco team led by Greg Marcus (first author: Eun-jeong Kim) was somewhat better than most and had a few notable twists.

The authors used the UK Biobank which includes about a half a million participants in the United Kingdom who underwent exams, provided blood samples, and were followed prospectively. Coffee consumption was the exposure of interest and incident tachyarrhythmia (atrial fibrillation [AF], flutter, ventricular tachycardia [VT], even premature atrial contractions [PACs] and premature ventricular contractions [PVCs]) was the outcome. They adjusted for confounders: race and ethnicity, educational level, body mass index, physical activity, hypertension, diabetes, and coronary artery disease (CAD).

One of the twists was that UK biobank participants also gave DNA samples, and this allowed the authors to construct a caffeine metabolism polygenic score by combining seven genome-wide significant variants (SNVs). A higher genetic score represents slower caffeine metabolism.

To validate these SNVs as genetic markers of coffee consumption, they tested the hypothesis that genetic makeup tracked with self-reported coffee consumption. And indeed, the gene variants influenced coffee consumption with slower metabolizers reporting lower intake. This allowed the researchers to do a Mendelian Randomization (MR) study—which has been called nature’s randomized controlled trial. Editorialists Zach Goldberger and Rod Howard called it a quasi-experiment. (The biggest MR I know of are the LDL genotypes in which having gene variants that associate with low LDL confers a very low risk of heart disease compared with not having the gene variants.)

Here is how I understand the MR in this study: the big problem with epidemiological studies like this is that self-reported low or high caffeine consumption may track with other important variables that are nonrandom, such as exercise or diet. But the gene variants occur at random, and if slow metabolizers (who drink less coffee because they get a bigger bang for their buck) had a lower event rate, you might have a clue that caffeine was causing that effect. The key with MR is that having or not having the gene score is naturally randomized.

Now to the two main results:

  • Over 4 years, and with adjustment for confounders, the authors found that coffee intake not only did not associate with an increase in any arrhythmia, but they also found that each additional cup of coffee was associated with a 3% lower risk of incident arrhythmia (hazard ratio [HR], 0.97; 95% CI, 0.96-0.98; P < .001).

  • In the Mendelian randomization portion, they found no significant association between underlying propensities of caffeine metabolism and risk of arrhythmia.

Taken alone, it is hard to draw conclusions from one observational study, but these findings add to the many contemporary studies that find no association between caffeine and arrhythmia. The absence of relationship between genetic markers of caffeine metabolism bolsters the conclusions that caffeine is not pro-arrhythmic, again, because the gene variants that effect coffee intake are distributed randomly.

I need to make a huge cautionary note on the finding that each additional dose of coffee was associated with a statistically significant lowering of arrhythmia. We should not take this as evidence that coffee is anti-arrhythmic. I say this because the sample size was huge, the effect size small at only 3%, and only a tiny fraction of participants reported arrhythmia. This is a classic problem with statistical significance in large databases: tiny effect sizes that are not clinically relevant can reach a statistical significance threshold.

In sum:

  • I don’t know that we needed another study that finds a lack of association between coffee and arrhythmia, but we have one and it is methodologically strong.

  • This paper makes the podcast because it forced me to review Mendelian randomization, which I find pretty darn cool; it bolsters tons of prior data; and by keeping coffee’s lack of harm in the news, it helps convert more of the public and doctors to the notion that one of life’s treasures ought not be feared.

  • That said, what I tell patients in the clinic bears repeating: while coffee doesn’t associate with arrhythmia in populations, by all means, if a patient feels more palpitations after coffee, they face a simple choice: drink the coffee and put up with the flutters or stop the coffee. Such is a classic preference-sensitive decision.

Final note: the other reason the non-connection between coffee and arrhythmia is important is to lessen distraction: people digest a finite amount of advice. Time and effort spent on caffeine is time and effort not spent on more important things: like not eating potato chips or ice cream or taking the anticoagulant.

CTA and Statin Use

JACC-Interventions has published an observational study done in Western Denmark studying the association between statin use and either no coronary artery disease (CAD) or non-obstructive disease in 33,000 patients who had scans for symptoms between 2008 and 2017. First author Kristian Overhus along with many colleagues made two comparison groups—about 20,000 who had no CAD and 13,000 who had non-obstructive disease (≤50%). They used national registries to determine statin use and cardiovascular events. Median follow-up was 3.5 years.

  • The main findings were that the absolute risk of the combined endpoint of myocardial infarction (MI) or death was directly associated with the CAD burden on the scan.

  • CAD burden was assessed with four groups of coronary artery calcium (CAC) score—zero, mild (0-99), moderate (CAC 100-399), and severe (CAC 400).

Remember, all these patients had either no CAD or non-obstructive disease. Also note that when I tell you the results, they are in rates per 1000 patient years.

  • Event rates were 4 in 1000 in no CAD; 7.8 per 1000 in mild; 13 per 1000 in moderate; and 32 per 1000 in severe.

That is not too surprising, but the second finding was:

  • Statin therapy was associated with an adjusted HR for MI and death of 0.52 in no, 0.44 in mild, 0.51 in moderate, and 0.52 in severe nonobstructive CAD.

That’s stunning because a 50% reduction in MI and death is way more than the expected 25% risk reduction seen in clinical trials. The authors tell us that in absolute terms, the risk reduction ranged from a number needed to treat (NNT) of ≈ 90 in no CAD to an amazing 13 in those with severe but non-obstructive disease. One important caveat: the authors did something important to sort out residual confounding; in other words, healthy patients took the statins and that is why they did better.

They did a negative control: they looked at statin use and pneumonia and cancer, two diseases that statins do not affect. There should be no association, but there was in this study: Table S9 in the supplement showed that statin use was associated with a 50% reduction in pneumonia in moderate non-obstructive CAD and a 36% reduction in cancer in those with no CAD. That strongly suggests residual confounding.

Table s8 in the supplement shows another finding consistent with confounding: statin use was associated with a 43% reduction in death in patients with no CAD. Now come on, even the most ardent statin proponent would not say that this drug slashes death that much in people with clean coronaries.

Another huge problem is that the authors don’t tell us what the pooled cohort equation would have predicted in these patients. We know the pre-computed tomographic angiography (CTA) probability of CAD is higher in those with disease vs no disease, but what I wonder about is how much more the imaging finding adds to the basic clinical data. For the typical intermediate risk person with a 7.5% 10-year risk, a scan might change your absolute event rate from 4 per 1000 to 32 per 1000, but how does this help the average person? In a grid of 1000 the difference between 4 and 32 looks terribly small.

People committed to doing everything to prevent an MI will take a statin regardless. For those on the fence, I don’t think a difference between 4 and 32 per 1000 will help sort in the decision to take a pill that doesn’t make them feel any better for a lifetime.

In sum:

  • This paper makes the podcast because CTA imaging is becoming increasingly popular for the exclusion of left main disease (at least that is how I have been using it) and I don’t think papers like this, with its serious confounding, can be used to support statin decisions.

  • I am surprised this one got by the zero CAC peer-reviewers. Zero CAC folks believe that no CAD, no CAC confers a risk so low that you can avoid statin therapy. But this paper reports a massive 50% risk reduction from statins in patients with no CAD. That defies all that we believe about statins and atherosclerosis.

Be careful using observational studies to overturn medical practice guided by oodles of RCT. It is old news that data from RCTs show that statins reduce event rates by about 25%. And 25% of a bigger baseline risk translates to a higher absolute risk reduction.

Appendage Closure

DRT stands for device-related thrombus and its translation is that a blood clot has formed in the arterial circulation in the left atrium (LA) on or near the foreign body that was implanted to prevent a blood clot from forming in the LA.

JACC has published a study from Mayo Clinic researchers that attempted to assess outcomes and predictors of DRT. My colleague and frequent debater Osama Wasni co-wrote the editorial—which is pretty darn good. Journalist Patrice Wendling has outstanding news coverage.

The Mayo-led group reached out to Watchman centers and asked them to send in cases of DRT, along with two cases at around the same time that did not have DRT. This allowed them to compare characteristics and attempt to put together predictors of those who had a clot and those who did not. They studied 711 patients total, 474 without DRT and 237 with DRT.

This study reported some important and sobering findings, but first its limitations. As Dr. Wasni and colleagues write in the editorial, the study cannot tell us the incidence of DRT because this isn’t a systematic collection of data, so we don’t know the denominator of total implants. Moreover, we already know the outcomes of a clot in the LA—it’s bad. It increases the risk of stroke by a manyfold. This study found a 3.5-fold greater risk of stroke, which is in line with prior studies.

But I see three other notable findings:

First, clots form on this foreign body late after implant; 39% occurred between 45 and 180 days, 16% between 6-12 months, and 20% occurred after a year. This is a huge problem. After the device is placed, the practice is to use an anticoagulant--warfarin or a direct oral anticoagulant (DOAC) short term and then as endothelialization occurs we are supposed to be able to stop the AC and switch to an anti-platelet drug like aspirin.

The whole darn point of this therapy is that its benefits are not in reduction of stroke (no trial ever has shown lower ischemic stroke), but elimination of the need of burdensome AC. But if you have a DRT or a clot in your heart you now must take AC. A DRT requiring AC totally, utterly, 100%, negates the benefits of having the thing in your heart. Remember, many of these patients are having the procedure because they (supposedly) could not take AC.

  • Second, the regimen of drugs that block clotting, so-called anti-thrombotics, which includes single antiplatelet therapy (SAPT), dual APT (DAPT), and AC did not associate with the risk of DRT. At the time of DRT diagnosis, most patients were being managed with either SAPT (36.3%) or DAPT (26.2%); 16 of the 237 or 7% cases of DRT were on oral ACs at the time of diagnosis.

This is strange because if you add in DAPT and oral AC, about one in three patients formed a clot despite hefty anti-thrombotic drugs. To me, this perfectly explains why the Watchman vs warfarin pivotal trials showed more ischemic strokes in the Watchman arm.

  • Third, the authors attempt to find predictors of DRT. This would theoretically help guide AC decisions post procedure. The biggest risk factor was any hypercoagulable state; second was post procedural pericardial effusion (of course if you have blood in the pericardium, you have to hold clot blocking drugs).

Chronic kidney disease and nonparoxysmal AF were also predictors of DRT. Again, these are the types of patients who receive Watchman. I don’t think we can make much of predictors because of the non-systematic data collection without randomization.

Watchman proponents will say, Mandrola, you’ve been in cardiology long enough to know that devices iterate and get better. You watch, we will design a better device and this problem will improve.

Maybe. But not likely in this case. Stents got a whole lot better than plain old balloon angioplasty and still, in patients with severe but stable CAD, there is clearly no reduction in MI or death. That is because atherosclerosis is a systemic disease and stents treat focal manifestations. Stroke is also a systemic disease, and LAAO is a focal solution for a systemic disease.

No matter the iteration, LAAO is likely to fail, especially when it is done in older patients with competing risks of stroke.


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.