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 30, 2021, John Mandrola, MD comments on the following news and features stories.
COVID-19
I am tired and frustrated and utterly confused about the new mask mandates and the move to do more testing. Do people not remember the 2 x 2 table and the problems with testing for low prevalence conditions? Hint: false positives.
You all know the story. It’s now a pandemic only for the unvaccinated, especially those with risk factors. I don’t envy policy makers. And I am moving on to a different topic.
Two Contrasting Papers on Alcohol
This week at the Heart Rhythm Society meeting in Boston, which is part virtual and part in-person, the famous group from Adelaide Australia presented an interesting study on alcohol intake and risk of atrial fibrillation (AF). JACC- Clinical Electrophysiology published the study. The group, with first author Samuel Tu, and senior author, Chris Wong, used data from the UK Biobank, which is a prospective cohort study of nearly a half-million middle-aged people from the United Kingdom. These are folks who were recruited and volunteered to have their health data followed over the years.
The research team studied the question of the impact of total and beverage-specific alcohol consumption on new AF. You may wonder why we need another alcohol and AF study. Many of Bradford Hill’s criteria of causation are already present. For example, listeners of this podcast know that there is biologic plausibility for alcohol causing AF, there are prior observational studies confirming a dose dependent association between alcohol and AF and a recent and elegant RCT showing that alcohol abstinence in patients with AF reduces AF episodes.
This study, however, addressed an important question that comes up all the time during the AF Clinic discussions of alcohol: Namely, can’t I have a little bit? Say, one drink per day? And herein lies the interesting aspect of this paper: compared with oodles of prior observational studies on alcohol and AF, the results had some similar findings and some novel ones. The similar findings include the observation that heavy drinking (of any beverage) was clearly associated with greater risk of AF.
The first novel finding was that at low consumption levels (less than seven UK drinks per week) there was a lower risk of AF. Then at levels of intake more than that, risk of AF goes up in a linear dose-dependent way. Researchers call this a J-shaped relationship. And it’s a novel finding because all previous observational studies have shown that any intake of alcohol increases the odds of AF. This includes a large recently published study from European authors using a similar design, a study of 100,000 middle-aged people in Northern Europe that found the classic linear dose-dependent increase in AF beginning with any intake of alcohol.
The second novel finding from the Australian team was that a beverage-specific analysis found that intake of beer/cider increased AF risk while red and white wine and spirits did not. When you look at the curves, AF risk goes up with any intake of beer/cider but with wine and spirits, the AF risk doesn’t increase until you get past 7 UK drinks per week.
These are observational findings. It is the only way to study this question. In a randomized trial, we would randomly assign one group to modest alcohol intake and the other group to higher doses, and we would completely control the dose of alcohol, diet, exercise, travel. You would have to imprison or impound humans for years to control everything. Obviously, this is not feasible.
As nicely laid out in the accompanying editorial from Thomas Dewland and Greg Marcus from the UC San Francisco, the key limitation in all observational studies like this is that participants chose to drink alcohol. Inherent in that choice are potfuls of other variables, only some of which can be adjusted for with stats.
Then there is the matter of recall bias—people self-report their alcohol consumption. The UK biobank also includes mostly White people who volunteer to be studied, which may produce a healthy-volunteer bias. This issue affects the generalizability of the data.
A couple of other notable facts: a UK drink (8 grams) is substantially smaller than a standard US drink (14 grams). So, the threshold here is actually 4 US drinks per week—a very small dose.
Another factor, clearly and laudably set out by the authors, was that they did not adjust for multiple testing. This means that the secondary findings on beverage-specificity have more susceptibility to false positive findings. They write that the associations “must be considered exploratory.” Whenever you read “these findings must be considered exploratory,” recognize that this is code for these findings may be due to the play of chance.
The strongest factor suggesting confounding on the matter of beer being worse than wine is that beer drinkers had more comorbidities.
How do we reconcile these findings with the previous observations showing no risk-free amount of alcohol? I don’t know but I suspect it is a matter of residual confounding. Or just a chance.
But if we simply put our common-sense hat on there are obvious messages:
Moderate levels of alcohol intake is a modifiable risk for AF.
High-levels are surely a risk.
Few electrophysiologists would posit that 4 drinks per week was problematic. But if a specific patient had alcohol as a trigger, it makes sense to reduce or stop it.
In patients with AF, reducing or eliminating alcohol intake stands a good chance of helping. That seems an important thing for patients to know.
This is the broader point: A patient recently asked me why there is so much more AF around. She had many friends with AF. Indeed, the AF epidemic does seem related to wealth: as nations become richer, its citizens eat more food, drink more alcohol, and feel more stress. The downstream effects of these are a perfect storm for AF. When the current crisis in communicable disease abates, we will be back to the paradox that despite our wealth and our medical advances and all this digital health, people aren’t living longer.
Second Alcohol Paper
The journal BMC Medicine has published a new analysis augmented with a meta-analysis of papers assessing alcohol consumption in relation to all-cause mortality and cardiovascular mortality in patients with known heart disease. The de-novo analysis included 14,000 patients enrolled in the UK Biobank study who had a previous myocardial infarction (MI), angina, or stroke. They then augmented these findings with a meta-analysis of 12 published studies on a similar matter and compared outcomes of drinkers against non-drinkers.
They found a J-shaped relationship with risk reductions, yes, risk reductions, with low to moderate-level drinking. And get this: No statistically significant elevated risks were found at higher levels of drinking. This led to a conclusion: People with established cardiovascular disease (CVD) might not need to quit drinking to prevent a repeat CVD event.
I am serious. This was published in a real journal. First author, Chengyi Ding, however, cautioned that the findings might "overestimate" the reduced risk for recurrent events for moderate drinkers due to the under-representation of heavy drinkers and the categorization of former drinkers, who may have quit drinking due to ill health, as nondrinkers in some of the datasets included in the analyses.
Journalist Megan Brooks, writing on theHeart.org | Medscape Cardiology, did something that all readers of health coverage should look for, and all journalists covering observational studies should do: She got a researcher to speak candidly about the flawed methods. She talked with Richard Saitz MD, a professor at Boston University Schools of Medicine and Public Health, who noted that in the few studies that excluded people who quit drinking because of illness, the effects of drinking were "attenuated."
"It is honestly tiresome that investigators continue to do these studies and journals continue to publish them even though they continue to have the same flaws," Saitz commented. "The bottom line" — based on work by others who used appropriate methods — is that alcohol in low amounts is "unlikely to provide cardiovascular or mortality benefits," Saitz added.
Indeed, the authors noted this finding but buried it in the supplement. (Always look at the supplement): “When studies with former drinkers in the reference group were excluded, the association was considerably weakened.” In RCTs, randomization (mostly) balances these issues so you can narrow down on the effect of the exposure (the drug, procedure etc). Another pro-tip: observational studies that report paradoxes (drinking is good for people with heart disease, or obesity is protective in AF or heart failure) are almost always due to bias—be it selection bias or collider bias. In other words, other factors.
The Stability of CAD
The journal Circulation has published a super important observational study of patients with stable coronary artery disease (CAD). This is a big one because it so perfectly confirms the findings of the RCTs, which have changed so few minds. The study is called CLARIFY, first author Jules Mesnier, senior author, Professor Phillipe Gabriel Steg. Remember that science tells us what we can do; trials tell us what we should do; registries tell us what we are doing.
CLARIFY is prospective registry of 32,000 patients with stable CAD (and yes, you had to really have strong evidence of CAD) who were enrolled in 45 countries—not the United States. The main findings are descriptive. I like it because it the authors don’t try to make causal conclusions about one group vs another. The study goal was to describe the prevalence and time course of angina as well as the effect of changes in angina status on outcomes. Follow-up was about 5 years. Table 1 shows that baseline medical therapy was excellent. More than 90% were on antiplatelets and lipid-lowering drugs and more than 70% were on beta blockers and ACEs or ARBs. About 7000 or 22% of enrolled patients reported angina at baseline.
You will never guess what happened to these patients with angina? Did they die off? Did they require PCI or CABG? No.
At one-year, in 40% the angina disappeared WITHOUT revascularization.
At five years, 84% of patients with angina had regression of it WITHOUT a new intervention.
Only 4.5% of patients with angina had to be controlled with coronary revascularization.
Patients whose angina had resolved at 1 year with conservative management were not at higher risk of cardiovascular death or MI than those who never experienced angina.
Not surprising, patients who developed new angina or had persistence of angina had a higher risk of CV events than patients without angina. In journalist Patrice Wendling’s excellent coverage, Senior author Prof Steg noted that, "It's perfectly reasonable to have a trial of watchful waiting with medical treatment in patients who have angina and stable CAD because the risk of events is low, the chance this will regress is very high, and if it regresses, they have an excellent outcome.”
Wendling writes that Steg, himself an interventionalist, said there is a financial incentive for revascularization in countries with some element of fee-for-service but, beyond that, there's also a true belief among interventionalists. "It's hard to believe that what you're doing is not useful, so you tend to want to have the vessel open and you tend to believe it is actually helping patients."
Also quoted is Christopher Granger from Duke: “This study adds to the evidence that medical management is very reasonable and may be the optimal treatment for most of these patients; and it's partly related to the somewhat surprising finding in this study that many of these patients had resolution of angina with just medical therapy.” And this: “Revascularization is still important for a portion of our population but it's way less important than we used to think it was."
I don’t know what it will take to change the clogged mindset of modern Cardiology. Perhaps it will take a generation or two. But the trials show it: BARI-2D, COURAGE, ISCHEMIA, all find that revascularization added to meds doesn’t reduce heart attacks or MI. And that is because stable CAD is so darn stable.
In this carefully done registry study, we learn than nearly 90% of patients with angina due to CAD experience regression of their symptoms. It gets better with meds. On. Its. Own. Not only is this important for doctors and patients to know, it is also important for scientists studying interventions.
What do I mean? This podcast has covered, over and over and over again the importance of placebo controls. After COURAGE and ISCHEMIA, the proponents of PCI say, Mandrola, you are heartless; PCI may not have reduced MIs or death, but angina was improved.
Yes, it was, and I don’t doubt it got better because patients in those trials knew their treatment assignment. If you got a stent or bypass, you clearly know it; if you did not get those things, you clearly know that too. So, did the angina get so much better after revascularization because it improved blood flow, or did it get better because of the placebo effect of feeling fixed?
This observational study showing frequent resolution of angina without intervention strongly argues for the placebo effect, and of course, it bolsters the results of one of the best studies ever done in Cardiology—the ORBITA trial. CAD is a focal manifestation of atherosclerosis. Atherosclerosis is a systemic disease. Medical therapy (statins, aspirin, lifestyle) work because they address the systemic disease.
Yes, in some cases, focal stenoses persist and patients benefit from revascularization, but CLARIFY confirms that it is the exception not the rule. Angina got better without intervention. Sleep on that.
© 2021 WebMD, LLC
Any views expressed above are the author's own and do not necessarily reflect the views of WebMD or Medscape.
Cite this: Jul 30, 2021 This Week in Cardiology Podcast - Medscape - Jul 30, 2021.
Comments