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, download the Medscape app or subscribe on Apple Podcasts, Spotify, or your preferred podcast provider. This podcast is intended for healthcare professionals only.
In This Week’s Podcast
For the week ending March 29, 2024, John Mandrola, MD, comments on the following news and features stories: the Intermittent fasting paper, complicated anticoagulation decisions, heterogenous treatment effects, frailty in heart failure (HF), the importance of the ECG, and industry conflicts.
The Intermittent Fasting Brouhaha
Dr Christopher Labos has a nice essay on theheart.org | Medscape Cardiology site about the recent intermittent fasting story. He is a great writer and thinker, but I have a bit of different take.
Perhaps you heard about this story. The American Heart Association (AHA) issued a press release in advance of the AHA EPI meeting detailing an observational study that found an association between people who self-reported intermittent fasting and a 91% higher rate of cardiovascular (CV) death. There was no simultaneous publication; it was just a poster.
This was about as flawed an analysis as there is. There were self-reported eating patterns (I can hardly recall what or how I ate yesterday, never mind weeks ago), and there were surely confounding variables — that is, people who self-report time-restricted eating may be different from those who do not. Dr Labos points out that the authors did 36 comparisons, so the play of chance was also likely. I could go on.
But this story has a special twist. One that really bugs me.
When the paper came out reporting that a popular pattern of eating was associated with harm, and mainstream media jumped on it, many of the Top People in medicine were outraged at the methodological flaws.
There was a backlash, and Twitter lit up with Top People telling us about the problems of non-random observational comparisons.
What bothers me about this is that many of these people remain silent when similarly flawed studies come out that find associations they like. Gulp, some of them even publish studies like the one they are publicly shredding.
My take is that nearly all of these sorts of studies are too flawed to have a) been done, b) been published, and c) been promoted or covered in the media.
What I try to do here is be neutral in my criticism of these studies. I don’t feel strongly about time-restricted eating, recent studies in the NEJM and JAMA find no real weight loss effects. What I would propose is that there be equal criticism for similarly flawed papers.
Today, I will discuss a number of observational studies that were done properly and add important findings to consider for the practice of clinical medicine.
Stroke Prevention with OAC in patients with AF
We have the simple idea that if stroke risk is high enough, we start oral anticoagulants (OACs). We determine stroke risk based on the simple CHADSVASC score. We estimate the untreated yearly stroke risk, then multiply it times 0.65, as 65% reduction is how much warfarin reduced stroke risk vs placebo or aspirin (ASA).
This is the prevailing thinking. It is what the guidelines say. It is the mainstay of decision aids for OACs.
But it’s super simple and makes assumptions that may not be true. One of the major assumptions is that there is no accounting for competing risk of death. Namely, if a patient dies of anything 1 or 2 or 3 years after starting OACs, there is blunting of the 5-year efficacy of stroke prevention.
The thing is, death is a common occurrence after a new diagnosis of atrial fibrillation (AF). Multiple observational studies report a 20% to 25% mortality in the first year after AF diagnosis.
Sachin Shah and his team at Massachusetts General Hospital had the idea to look at how these competing risks affect benefit of OAC. Circulation Outcomes published their study. They chose a neat way to explore this question:
They had data from 12 warfarin vs placebo or ASA trials. The warfarin trials were the only OAC trials that studied AC vs no AC. Recall that the direct acting OAC trials studied direct acting OACs vs AC with warfarin.
For each person in the trials, the authors estimated an absolute risk reduction (ARR) of warfarin with two methods. One was to use the guideline-endorsed model — 0.65 times risk. This was simple. Each patient gets a no-treatment stroke rate based on the Friberg and colleagues’ Swedish cohort study. They then extrapolated this risk reduction to 5 years.
The other method was to use a competing model that used the same CHADSVASC score but also accounted for competing risk of death. To determine the risk of death, they used life tables. This gets a little complicated b/c it involves something called the Fine-Gray extension of the proportional hazard model, which treats death from non-stroke causes as a competing event. Basically, the older and sicker a patient was, the greater chance that non-stroke death would happen.
The best way to understand this is to picture a graph with years (or time) on the x-axis and ARR of stroke on the y-axis. When you use the simple 0.65 x the untreated stroke rate. It’s linear because the yearly risk reduction just increases by the same amount over years. By 5 years, there is a 10% ARR.
But with the competing risk model, the amount of risk reduction does not keep going up. It’s attenuated after a year or two because you can’t benefit from stroke reduction with AC if you die of something else.
This was the first big finding: the simple CHADSVASC estimation of ARR overestimates the benefit compared with a model that accounts for death. The actual numbers were as such: ARR with AC with CHADSVASC model at 3 years was 6.9% vs 5.2% with the competing risk model (P < 0.001).
They then looked at how life expectancy affected the difference between the simple CHADSVASC model and the competing risk model. If your life expectancy was short, say 1to 4 years, the CHADSVASC model overestimated the benefit by nearly 80%. But if your life expectancy was long (>11 up to 47 years), the competing risk model underestimated the benefit compared with the simple model.
The bottom line: They showed that while warfarin was clearly effective, treatment benefit was overstated when using the guideline-endorsed approach because guidelines do not account for the competing risk of death and assume a constant growth in treatment benefit over time. Overestimation was most pronounced in patients with the lowest life expectancy and when the benefit was estimated over a multiyear horizon.
Comments. I find this a super-clever and super important analysis. Like you, I see patients who have advanced age and multiple risk factors for competing risks of death. The older patient has a high CHADSVASC. We jump to OAC because the stroke risk is high. Conversely, we see a 55-year-old with hypertension and think that’s pretty low risk, and we may hold off on OAC.
What this analysis does is to make us think that younger patients with AF may live longer, that is, long enough to benefit more from stroke prevention than the older higher risk patient.
The data is not enough to act on, specifically. But it’s enough to make us think that simple guideline-endorsed recommendations may not fit the older person who has substantial competing risks of death.
This concept is similar to why I am increasingly thinking that primary prevention therapies may deliver more benefit to younger lower-risk people, because they have longer to benefit.
One important caveat, one that I learned first from Dr Jeremy Sussman at University of Michigan, on Twitter. Dr. Sussman reminded me that, in the same way that OAC-benefit is reduced in a competing risk-of-death model, so are bleeding rates. In other words, if an early death effectively reduces OAC benefit, early death would also, theoretically, reduce bleeding risk. Because if you die, you don’t have a bleed.
I discuss this all because competing risks don’t get enough attention.
This is another reason why I don’t love those simple colored boxes in guidelines. Even the decision to use OAC is complicated and turns on the characteristics of each patient, as well as, of course, his or her preferences.
Another Paper on Co-Morbidity and Treatment Effect
My buddy Andrew Foy and his team at Penn State have another analysis that expands on the competing risk of death analysis I just told you about. They did a re-analysis of eight CV randomized controlled trials (RCTs) exploring the relationship of co-morbidity to treatment effect. Three ACCORD trials, AFFIRM, BARI-revascularization, SCD-HEFT, SPRINT, and 2 TIMI-trials.
Since they had source data from these eight trials, they could apply the Charlson Comorbidity index (CCI), which is a way of quantifying the degree of co-morbidity, to see how that affected treatment effect. Trialists do that with subgroups but a subgroup is just one factor. The CCI accounts for lots of things.
They found that most trials enroll patients with low CCI and little variability of the CCI. We know that, but they showed it nicely.
Their second finding was that about one-third of trials had clinically relevant interactions between the CCI and treatment effect.
For instance, in SCD-HeFT, the majority of treatment benefit was in patients with low CCL. In AFFIRM, rate vs rhythm in AF, which found a non-significant 15% higher rate of death in the rhythm control arm, there was clear harm from rhythm control in those with high CCL.
Foy’s paper is an eye-opener.
In their discussion, they point out why the interactions could have occurred. Usually, the explanation is similar to the competing risks of AC study; namely, that you have to consider not only the probability of treatment benefit, but you also have to consider the probability of harm, the probability of having the primary outcome event as well as the probability of a competing event. This amalgam of stuff is what clinicians do. It’s why robots can’t practice medicine.
I love the idea of having to consider all of these complex probabilities. It is exactly why clinic days are harder than days when all we do is isolate pulmonary veins in the electrophysiology (EP) lab.
Frailty HF and Cause of Death
Circulation Outcomes published another important paper on competing risks. In this one, a Japanese group reported data from a registry called the FRAGILE HF cohort comprising older patients, who were hospitalized for HF in 15 hospitals in Japan.
Their goal was to study the association between multidomain frailty and cause of death. Follow-up of this study was 2 years.
They first set out three causes of death a) HF death; b) other CV death (say coronary, sudden cardiac death (SCD), stroke, RF; and c) non-CV death.
They made three categories or domains of frailty — one physical, one social, and one cognitive.
The total cohort was about 1200 patients; mean age 81. Stop there. This is the most common age of HF patients I see at a busy community hospital.
45% of patients had 0-1 domains of frailty.
37% had 2 frailty domains.
18% had 3 frailty domains.
The main findings were:
A greater number of frailty domains (FDs) were associated with a higher rate of all-cause death.
As the number of FD increased, the prevalence of SCD, stroke death, death due to noncardiovascular causes, and unknown death numerically increased, whereas that of death due to HF decreased.
Only non-CV death and non-HF death and other cardiovascular death were significantly associated with the number of FDs.
In a competing risk analysis with left ventricular ejection fraction (LVEF) as an adjustment variable in addition to age and sex ,they found that HF with preserved EF (HFpEF) was more prevalent in those with a greater number of FDs.
Comments. I highlight this paper, not only because it is a good use of observational data, but because it stays with the theme this week of thinking about the complexity of patients with HF. Guidelines want us to put these patients into boxes — HF with reduced EF, give the big four drugs, and HFpEF start sacubitril/valsartan and SGLT2 inhibitors, but many of these patients have HF and other things.
Cardiologists aren’t trained in frailty; I have never used a frailty score, but I don’t think that matters. When we walk into a hospital room, we ought to be able to recognize it. Ask what this patient does — physically, socially, cognitively. Ask yourself: would this patient have been randomized into one of the seminal trials that included mostly stable ambulatory outpatients?
If your patient has one or more of these frailty domains, it’s likely that they face a non-CV cause of death. Congratulations to Circulation Outcomes for publishing such nice papers, and to the Japanese group, first author, Koichi Ohashi.
The Importance of the 12-Lead ECG in Choosing EP Procedures
I went into cardiology because of the pure beauty of the ECG. A group from the University of Chicago (UC), publishing in JAMA-Cardiology has a useful paper on the importance of notches on the ECG.
The background goes like this: left bundle branch block (LBBB) is a problem for patients with HF because when the LBB is blocked, activation of the right ventricle (RV; down the right bundle) happens first. Activation then causes contraction so we get the RV contracting first. Then muscle-to-muscle conduction from right to left ventricle occurs later. This causes LV contraction after RV contraction. On echo, it looks like a swinging or bouncing heart.
The dys-synchrony between ventricles can be terrible as it can cause or worsen LV function and exacerbate heart failure.
Cardiac resynchronization therapy (CRT) is 100% electrical therapy. By placing pacing leads in the RV and LV (the epicardium accessed via the coronary sinus), you can pace both ventricles simultaneously. The biV pacing synchronizes the ventricles.
On the ECG, one feature of LBBB is that it is wider than normal. I will come back to that as it is at the core of the UC authors’ work.
Conduction system pacing via the His bundle pacing or left bundle area pacing also treats the RV-LV dys-synchrony by directly capturing the conduction system which then leads to simultaneous activation of both ventricles.
But here is the rub. The QRS can be wide and LBBB-appearing on the ECG and it is not due to delay in the conduction system. This happens when the problem is delayed conduction through the myocardium or muscle. We call this an interventricular conduction delay (IVCD). The conduction system is fine, but the muscle is diseased, and conduction proceeds slowly through the diseased muscle and the QRS is still wide.
Conduction system pacing (CSP) will not work for this problem. Because CSP can only correct a block in the conduction system. The muscle-to-muscle conduction is downstream from the conduction system. It’s also hard for standard biV pacing to correct this issue as well.
IVCD occurs often in patients with heart failure, LV hypertrophy, or many other causes.
Therefore, what we have to do as cardiologists is decide whether the wide QRS is caused by bad muscle conduction or conduction system disease.
Enter the UC group which has previously done elegant recordings of the left side of the septum wherein they can show the site of block in the LBB. Their early papers were led by Dr Rod Tung. This new paper, which has absolutely gorgeous images, harnessed the value of those LB recordings to help sort out a new ECG criteria for LBBB.
Here is a brief summary of their recent work:
They had about 75 patients who, on baseline ECG had a LBBB-appearing QRS pattern, who had these LB recordings done during an EP study for something else.
A little more than half had a true block in the LBB and the others had normal conduction. Right there, you know that not every LBBB-appearing QRS has true block in the conduction system.
Then they did correlations looking for the most predictive ECG patterns that would identify (sensitivity) the true conduction block, and correctly identify the non-conduction block IVCD cases (specificity).
They found a simple measure that worked best. It was a time to notch in lead I of greater than 75 millisec. That’s nice because that is an easy measure. The sensitivity and specificity were not perfect, but the specificity — probably the more important value, because we don’t want to implant devices in patients who won’t benefit — was far better than the standard criteria, called the Strauss criteria.
But they were not done. That they called the derivation cohort. Cleverly they came up with a validation cohort. For this they used patients post-transcatheter aortic valve implantation (TAVI) who acquired a new LBBB. This was clever because post-TAVI LBBB is mostly always related to impingement on the LBB (ie, true conduction block).
Here the sensitivity was 87%. But there were also 10 of these post-TAVI-LBBB patients who had a pre-procedural IVCD that looked like an LBBB. In these patients, the time-to-notch-of-greater-than 75 millisec correctly identified IVCD from true LBB in all cases. In other words, 100% specificity.
Comments. This is brilliant work that is not only methodologically sound but clinically relevant. We as clinicians need to be able to identify true conduction block vs IVCD on the ECG. For this, we do not need artificial intelligence, genetic scores, or advanced imaging. We simply need to understand the heart’s anatomy and physiology. And be skilled in reading one of cardiology’s most important tools — the inexpensive and no-risk 12-lead ECG.
The UC authors have helped us better distinguish true conduction block. This is important, as it informs common decision about the type of electrical therapy to consider in patients with heart failure.
I say congratulations to them.
I also want to compare this sort of foundational electrical work with the ablation side of EP. I worry about the ablation side. Here, instead of doing foundational work, we seemed enamored by, perhaps distracted by, better ways to destroy left atrial (LA) myocardium. My Google alerts and my Twitter feed overflow with marketing for different ways to obliterate LA cells, in the name of capturing more of the AF ablation market.
My hope for the AF side of EP is to take a cue from the foundational work done in CRT therapies such as that done by the UC team.
I’ve been ablating AF for 20 years, and we still have not a clue what causes this condition. We have no clue about the best strategies to ablate AF, we have no proper sham-controlled trials of AF ablation.
What we have now are $9000 ablation catheters rather than $3000 ablation catheters. That is hardly progress.
Industry Payments to US Physicians by Specialty and Product Type
Speaking of Industry, may I point you to a brief JAMA Research Letter that I co-authored this week. Full credit for this work needs to go to Ahmed Sayed, an aspiring young doctor in Cairo, Egypt, who plans a career in cardiology. Co-authors of this paper included major academics, Joe Ross, Lisa Lehmann, and Andrew Foy.
Ahmed was a force of nature doing this work and the many revisions. The paper took data from the Open Payments platform from 2013 to 2022. We only included cash and noncash equivalents. Royalties and research funds were excluded.
Finding one: Industry paid out $12 billion to US docs.
Finding two: About 57% received at least one payment from industry. (so there are actually lots of non-conflicted docs).
Finding three: Payments varied widely across specialties. The highest payments went to Ortho (~$1.4 billion), Neuropsych, and Cardio (~$1.3 billion each.)
Finding four: Across all specialties, the average (median) physician received only $10s to $100s, whereas top earners often received > $1 million.
Finding five: We showed the 25 medical products with the most payments separated by drugs and devices.
Top drugs that involved payments – rivaroxaban and apixaban were one and two. Hundreds of millions of dollars. Numbers 5 and 6 on the list, empaglifozin and dapagliflozin. You know what wasn’t on the list: spironolactone. Metoprolol.
The top device, by a ton, was the daVinci Surgical System. Cardiac devices included, wait for it, the EVOLUT AV, Impella, LIFEVEST, and Watchman.
You know what wasn’t on the list: The 3830-pacing lead, used for conduction system pacing.
Comments. This research letter will not solve the problem of profit-driven medicine. I am not against profits, as profits can lead to innovation. But the sheer amount of money going to doctors is astounding. And it makes me think. Do take a look at the JAMA publication. It’s super short. Let me know what you think.
Next Week
Lancet published a paper on diabetes risk with statins. The University of Pennsylvania group published a super interesting paper on using type IC antiarrhythmic drugs to suppress PVCs in, get this, patients with advanced heart disease.
© 2024 WebMD, LLC
Any views expressed above are the author's own and do not necessarily reflect the views of WebMD or Medscape.
Cite this: Mar 29, 2024 This Week in Cardiology Podcast - Medscape - Mar 29, 2024.
Comments