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 March 26, 2021, John Mandrola, MD comments on the following news and features stories.
Mixed News on COVID This Week
First the bad news: My friends in France, Poland, Germany and Italy tell me of increasing hospital admissions. When I look at very low vaccine numbers in the European Union relative to the United Kingdom and the United States, I wonder why that would be? In the United States, Michigan, New York, and New Jersey are trending up, which is weird, because many states are seeing continued and striking declines in cases. I would never have thought places like Kentucky and West Virginia would ever lead any indices in health matters, but we are doing very well.
Some good news: Despite some SNAFUs with the press release, the Astra Zeneca vaccine looks quite good. Efficacy in the high 70% range, but more importantly, zero vs eight severe cases in the vaccine vs placebo groups, respectively. Now we have four vaccines.
I saw a picture this week from Tel Aviv showing a line of shops and cafes with people acting like people, namely, they were sitting and talking together—and without masks.
Finally, a research letter in JAMA reported on neutralizing-antibody response to four variants in infected and vaccinated individuals. The goal of the Emory-led team was to determine how mutations within the spike protein are associated with virus neutralization. The great news is that they found “neutralizing activity of infection- and vaccine-elicited antibodies against four SARS-CoV-2 variants, including B.1, B.1.1.7, and N501Y.”
AF Detection After Stroke
Many years ago, I heard Stanford Professor John Ioannides remark that doctors and scientists were industry’s best marketers. He showed examples of how industry disguises marketing as science. A study presented at the International Stroke Conference beautifully exemplifies this concept.
The Stroke AF study is an industry-sponsored study comparing the use of a pricey implantable cardiac monitor (ICM) to standard of care in patients who have had a recent stroke. Enrolled patients had an ischemic stroke “believed to be due to small vessel, large vessel, or intracranial atherosclerosis.” Cardioembolic strokes were excluded. The study’s primary endpoint was detecting more than 30 seconds of atrial fibrillation (AF) during the year after the stroke. The mean age of enrolled patients was 67 years and the mean CHADSVASC was 5.
To no person’s surprise, AF was detected in 12% of those with ICM vs 2% in the control arm. About 44% of patients experienced an episode that lasted more than 4 hours. The median time to first detection of AF was about 3 months. Oral anticoagulants were prescribed more often in the ICM arm.
These findings led the principal investigator to say, “Thirty days of ambulatory cardiac monitoring, which is a common practice in the United States, even among patients with cryptogenic stroke, would have been insufficient to capture the vast majority of AF episodes that occurred in this trial.... Clearly, use of ICM in this population may be beneficial to detect poststroke AF and to inform optimal stroke prevention strategies."
This trial began enrolling in 2020, 4 years after the CRYSTAL AF study showed nearly the exact same thing in patients with stroke of unknown source. Eight years before that, the ASSERT study found that if you have monitors (in this case pacers/ICDs) in patients with an elevated CHADVASCs score, you pick up AF in about 10% of people.
In the news story from the stroke conference, the cited outside expert—a professor—noted STROKE-AF’s biggest limitation was the lack of control group. This is weird because the control arm was standard of care, meaning monitoring was up to the treating doctor. Thus, the study did have a control arm.
My take of this study’s biggest limitation is that it doesn’t tell us anything useful. We have oodles of data that shows we will pick up incident AF in older patients with risk factors. We have two trials showing that direct oral anticoagulation after stroke of unknown source does not reduce stroke more than aspirin (NAVIGATE-ESUS and RE-SPECT ESUS). To me, this was a post-market-release study that had an obvious outcome, and provided no relevant information to patients or clinicians.
Preventing stroke after a first stroke is an important thing to do. For this we need studies powered to measure important clinical outcomes, not surrogate outcomes like 30 seconds of AF.
Measuring Heart Function
JAMA-Cardiology has published a new study looking at the predictive ability of subclinical measures of heart function. Right off the bat you should be alert to the term subclinical, which is medical jargon usually reserved for patients without complaints who are usually not asking for our help, yet.
The paper’s authors, first author Anne Marie Reimer Jensen, assessed the value of adding strain measures to the left ventricular ejection fraction (LVEF). Strain is deformability of the LV. Too little is bad. Consider that 20-year-olds are more deformable than 80-year-olds. Software in echocardiography machines has advanced to the point where we can measure strain as longitudinal and circumferential, although there are lots of technical caveats.
Let’s first address the LVEF. Regular listeners well know that the main problem with current-day measures of the LVEF is its massive over-quantification. In truth it is abomination to call an EF 36%. Or 48%. Or 29%. The EF varies a ton and giving it one number is one of cardiology’s biggest jokes.
Doctors will say the EF is load-dependent, but this just means that if the patient is a bit dry or a bit wet, fluid-wise, the EF may vary. The EF may also vary depending on the presence of ectopy or heart rate. But even crazier is that EF can vary depending on where on a shadow the sonographer puts the caliper. That may vary within or between people. EF would be fine if it was reported, as Professor Feigenbaum told us at Indiana University in the 1990s, in four categories—normal, mildly, moderately, and severely impaired.
Increasingly, my echo colleagues are telling me that my patient has abnormal strain. That is why I was drawn to the current study. I thought, perhaps this paper is going to help me sort out what I should think when my patient’s echo shows abnormal strain.
The investigators used the Atherosclerosis Risk in Communities (ARIC) database to assess the independent associations of subclinical impairments in systolic performance with incident Heart Failure (HF) over 5 years. ARIC is one of those community-based prospective cohort studies in which adults are followed over time.
The exposure variables were LVEF, longitudinal strain (LS), and circumferential strain (CS) measured at the fifth study visit. The outcome of interest was HF of any type over the next 5 years. Among nearly 5000 ARIC participants about 3500 had complete assessments of EF, LS, CS.
The Hazard Ratio (HR) per one standard deviation (SD) decrease in EF was 1.41 (1.3-1.5)
The HR for LVEF < 60% was 2.59 (1.9-3.6)
The HR for one SD decrease in LS was similar -about 1.30 with a similar confidence interval
The HR for one SD decrease in CS was also 1.39
In sum, using current cutoffs for normal EF of 54% in women and 52% in men was less predictive than less than using 60%. Adding strain measures may also increase prediction of future heart failure.
In an accompanying editorial, two HF experts, Dr. Clyde Yancy and Greg Fonarow, wrote that the implications of this work were profound. I asked Dr. Fonarow on Twitter to expand how measuring strain will help. He said, “By identifying patients at increased risk for incident HF, inclusive of this older age range, we can intervene with HF preventative therapies/risk factors interventions”
Now let me tell you about who these patients were:
Median age: 75 years.
81% with hypertension.
63% on antihypertensive meds.
32% with obesity.
You can see the problem. Just being 75 years old is a risk factor for heart failure over the next 5 years. Dr. Fonarow told me that knowing that a patient’s EF is below 60% or that her strain is 1 SD too low may help me initiate therapy to help prevent future HF. I completely understand that notion. But what might these therapies be?
Exercise? I’d recommend that regardless of the strain.
Good diet? I’d recommend that regardless of the strain.
BP meds? Same.
Diabetes? I’d be looking for a reason to use an SGLT2 inhibitor, again, regardless of the strain.
Obesity? This podcast recently had a positive take on semaglutide. But would a strain measurement or EF of 59% influence this decision?
Now let’s consider what the harms of this could be. A well-meaning doctor reads the paper and editorial and then thinks, Gosh, my patient has both an EF less than 60 and abnormal strain. I’d better add a diuretic to the regimen. Boom, one-month later, the patient is admitted for confusion due to a sodium of 125.
I don’t mean to sound like the old doctor who is close-minded to new stuff, but this paper does not convince me that our current paradigm is broken. Strain may have a role in the future. But we must balance the quest for perfectibility against the potential tradeoffs.
This reminds me of the book I am reading now by Thomas Sowell – The Conflict of Visions. Sowell speaks of the constrained vs unconstrained visions. It’s a book mostly about public policy but Sowell’s visions apply well to one’s approach to medicine. Both camps want good outcomes; but the constrained vision accepts imperfection and feels that there are no perfect solutions, only tradeoffs. The unconstrained vision holds that, with enough effort, led by smart people, there can be great solutions.
The quest to perfect LV function measurement, or define coronary calcium, or screen for specific diseases, are examples of unconstrained thinking. In other words, with enough effort, we can find solutions to human suffering. I worry about medicalization and indication creep. I worry about the tradeoffs.
A promising use of strain may be in cardio-oncology. The unconstrained vision would have the detection of early problems from chemotherapy as a great gain—with enough effort, we will reduce cardiac complications of cancer therapy. But the constrained vision worries that a risk-averse cardiologist sees this strain imaging and recommends altering cancer-therapy. That could go either way—good or bad.
Maybe I am naïve, but it seems the conflict of these visions—in Medicine, at least--must be resolved with empirical study. We just need the courage to recognize tradeoffs. Medicine is replete with examples of things that made sense but ended up not being of benefit—due often to tradeoffs. Think, for instance, of all that data a Swan-Ganz catheter gave us in the ICU. So I would say, if you want me to change the paradigm of caring for people using new measures of LV function, study it and show me that is better, or at least not worse.
Pay for Performance
Speaking of constrained and unconstrained visions, let’s apply the two philosophies to the problem of varying quality of healthcare in hospitals. Variation is a problem because it implies that some places are doing good work and others are doing inferior work.
Everyone agrees that the goal is better quality in all hospitals. The constrained vision would accept that there will be variation—because humans vary—and strive for the program of improvement with the most favorable tradeoffs. The unconstrained vision believes that with the absolute smartest policy, a solution could be had that raises all hospitals to a high level with low variation.
In the United States, government payers, eg the Center for Medicare and Medicaid Services, have instituted policies characterized by a stick rather than a carrot–they have tended more to unconstrained thinking. Taken together, policies with long names like Hospital Value-Based Purchasing Program (HVBP) and Hospital Readmission Reduction Program (HRRP) are pay-for-performance measures that create penalties and bonuses for hospitals that show improvements in outcomes like risk-adjusted reductions in 30-day readmissions for specific conditions.
JAMA has published a super-important research letter, the Association Between the Proportion of Black Patients Cared for at Hospitals and Financial Penalties Under Value-Based Payment Programs. The authors are well-known policy researchers form Beth Israel in Boston and Washington University in St Louis. The research question is whether these initiatives disparately affect hospitals that care for a high proportion of Black patients. Translation: is there a tradeoff? The key result: After adjustment for hospital characteristics, including being a safety-net hospital, high-proportion Black hospitals were more likely than other hospitals to be penalized and less likely to receive a bonus.
Senior author Rishi Wadhera has a nice thread on Twitter. Wadhera noted that already a large body of evidence suggests these pay-for-performance programs have not meaningfully improved hard outcomes. But now this paper suggests these programs contribute to structural racism by disproportionally levying financial penalties on already under-resourced hospitals. Crucially, Wadhera notes these deleterious structural effects are totally unintentional.
But that’s the thing that highlights the tension between constrained and unconstrained visions. The constrained vision holds that there are no perfect interventions. Tradeoffs must be considered. I ask you to imagine a world in which pay-for-performance was studied in pragmatic trials before being implemented. We would have likely known about these problems and lack of effect on clinical outcomes before exposing millions of people to them.
Smart people making policy with the best of intentions are not enough. Empiricism must also apply to policy.
© 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: Mar 26, 2021 This Week in Cardiology Podcast - Medscape - Mar 26, 2021.