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 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 June 10, 2022, John Mandrola, MD comments on the following news and features stories.
More ISCHEMIA Controversy
ISCHEMIA Substudy Data Just Don't Add Up, Cardiac Surgeons Say
There is even more controversy surrounding the Reynolds and colleagues substudy of ISCHEMIA published in Circulation in 2021. Patrice Wendling had the story and it came out Friday of last week just after I finished recording.
I cover it here as the lead because ISCHEMIA was a 100-million-dollar National Institutes of Health (NIH) funded trial that corroborated the COURAGE and BARI-2D studies — that is, the notion that patients with stable coronary artery disease (CAD), even severe multivessel disease, with ischemia, could be treated with medical therapy first and only proceed to angiography and revascularization if symptoms persisted.
The New England Journal of Medicine (NEJM) published the main ISCHEMIA trial in April of 2020. The title said it all: “Initial Invasive or Conservative Strategy for Stable Coronary Disease.“
More than 5000 patients were randomized.
There were no significant differences in the composite primary outcome of cardiovascular death (CVD), myocardial infarction (MI), unstable angina (UA), heart failure (HF), or cardiac arrest.
The delayed strategy resulted in similar outcomes with far, far fewer coronary angiographies, stents, and bypass surgeries (CABG).
The message was that you could take a patient with a positive stress test and treat them medically.
Now, when there is a 100-million-dollar trial with more than 5000 patients, you can expect sub-analyses, looking into subgroups and outcomes in certain groups. COURAGE, for instance, has more than a dozen substudies.
Substudies can be interesting. They can help you form questions for future trials. But you have to be cautious about making strong conclusions because the main trial is statistically powered to reasonably avoid confusing signal from noise. When you start looking at fractions of the main group, the risk of noise goes way up. The other problem with substudies is that they often ask different questions than the initial randomization. When you do that, it becomes an observational study, and that comes with all the biases of non-random comparisons.
One terrible conclusion that likely came from looking at a subset of patients of ISCHEMIA was the guideline recommendation to downgrade CABG in patients with three vessel disease from a I (recommended) to a IIb (may be considered). This likely came from the fact that revascularization in the invasive arm of ISCHEMIA could be by percutaneous coronary intervention (PCI) or CABG. The vast majority of patients in ISCHEMIA had PCI. Only a small fraction of patients had CABG. So, the surgeons are rightly upset by this downgrade. It is wrong to extrapolate the main findings of a 5000-patient trial to a subset of only a hundred or so patients.
Now to the Circulation substudy by Reynolds and colleagues, published in September 2021. Here they looked at the relationship between anatomic severity of CAD and degree of ischemia and outcomes, overall and by management strategy.
They could look at anatomic severity because about half the group of patients could be classified using the modified Duke criteria — a way to rank the ‘badness’ of CAD. An inclusion criteria was ischemia, so the degree of ischemia could be known in nearly all patients. Before I tell you the results, remember that this substudy is actually an observational study within an randomized controlled trial (RCT).
Reynolds and colleagues found that:
Relative to mild/no ischemia, neither moderate ischemia nor severe ischemia was associated with increased mortality.
However, anatomic severity of CAD did correlate with higher death rates.
An invasive strategy did not lower all-cause mortality in any ischemia or CAD subgroup.
You can see how these conclusions would also support a downgrade of CABG, right? The thinking goes that no matter the degree of ischemia or CAD severity, early invasive revascularization did not make a difference.
But this is a problematic conclusion –there are oodles of studies correlating the degree of ischemia with a higher risk of death. RCTs are for comparing outcomes of two treatments. Predicting outcomes based on a variable such as degree of ischemia seems better suited for larger observational cohorts.
But the other problem was that the initial manuscript had data inconsistencies from the main paper compared with the supplement tables. The two surgeons pointed this out in a letter to the editor, that was not published. A formatting glitch in the dataset explained this issue.
But the two surgeons looked further and found more inconsistencies, now between the main trial in NEJM and this substudy. This was detailed in a second letter (agree to terms of service to download the letter), which Circulation also refused to publish.
I won’t detail all the specific issues here, but the surgeons wrote that the multiple discrepancies in numbers and percentages could not be explained even after adjusting for different denominators or missing values. A formatting error was implausible.
But there is more. Patrice Wendling did an incredible job covering this story:
...The lone cardiac surgeon on the ISCHEMIA trial steering committee, T. Bruce Ferguson, MD, has resigned from the committee, citing a series of factors including an inability to reconcile data in the substudy and two additional ISCHEMIA papers currently under review.
Ferguson, now with the medical device manufacturer Perfusio Corp., said he resigned from the ISCHEMIA steering committee on May 8 after being unable to accurately reconcile the ISCHEMIA surgical subset data with the Reynolds substudy and two other ISCHEMIA papers on the CABG subset.
At least one of those papers, he noted, was being hurriedly pushed through the review process to counter concerns raised by surgeons regarding interpretation of ISCHEMIA.
"This is the first time in my lengthy career in medicine where a level of political agendaism was actually driving the truck," he said. "It was appalling to me, and I would have said that if I was an interventional cardiologist looking at the results."
"Something political between 2020 and 2022 has crept into the ISCHEMIA trial mindset gestalt and I don't like it," Ferguson said. "And this can have enormous consequences."
The ISCHEMIA authors are working on a second correction. Circulation did not publish the second letter, but it is available online
Comments. The Ferguson comments are striking, even for someone leaving academia. CABG recommendations should not change based on a tiny subset of ISCHEMIA. The discrepancies and the failure of Circulation to publish these letters worries me.
All this may be innocent, I have no idea about any inside story related to Ferguson, but clearly science has a trust problem and stuff like this does not help. I hope there is a reasonable explanation. Please let there be a reasonable explanation. I’ve said this before but will say it again: practice-influencing trials should have to release their raw data for independent review. I strongly believe that main trial results should be the evidence that influences practice. Substudies should be interpreted with great caution.
Health Insurance and Health Outcomes
JAMA-Cardiology has published a super-interesting paper from the Beth Israel group in Boston – first author Andew Oseran and senior author Rishi Wadhera — on how insurance coverage affects health care access and CV risk factor management for working-age adults with low income.
This topic may sound weird to some of our international listeners who work in countries with public health insurance or public health care. But here in the United States, healthcare insurance most often stems from employment by a big company. If you are self-employed or work for a small company, you have to buy insurance privately, and it is super expensive.
Americans who are sufficiently poor may qualify for government funded insurance, called Medicaid, which isn’t great but does grant access to basic health care to those with low- or no-income. We have a debate here about how much Medicaid should be offered. When Congress passed the Affordable Care Act years ago, some states expanded Medicaid; some states did not. This set up a quasi-randomized comparison. This nifty study correlated outcomes based on the exposure of Medicaid expansion or not. The main outcomes were health care access and the treatment of cardiovascular risk factors. Buckle up for this one.
The study comprised 28 million working age adults; about one-third were in non-expansion states and two-thirds were in expansion states. If you were a low-income adult in a nonexpansion state you:
Had lower rates of insurance;
Were less likely to have a regular doctor;
Were less likely to have had a recent exam;
Were more likely to have deferred care because of cost.
This all has to be bad, right? Having basic healthcare is important. But the only outcome of this study was the treatment of CV risk factors. Arguably an important thing. Sit down for this: There were no significant differences in CV risk factor management between these groups.
Comments. This study is not an outlier. There have been three major trials that have all found that more health insurance leads to more healthcare but has no appreciable effect on outcomes.
The RAND Health Insurance Experiment included 7700 people in six US cites, randomly assigned to either free or not free healthcare. The group given free care consumed 30% to 40% more services but did not have better outcomes in most metrics.
A few years later came the Oregon Health Insurance Experiment. Oregon assigned a limited number of available Medicaid slots by lottery. Those who won the lottery and got Medicaid had substantially and significantly better self-reported health but there was no statistically significant effect on measures of blood pressure, cholesterol, or blood sugar. And more insurance did not reduce the predicted risk of a cardiovascular event.
The third study was the Karnataka Hospital Insurance Experiment; in-patient coverage only. The sample included more than 10,000 households in 435 villages in India. Sample households lacked other [hospital] insurance and were randomly assigned to one of four treatments: free insurance, the opportunity to buy insurance, the opportunity to buy plus an unconditional cash transfer equal to the premium, and no intervention.
The opportunity to purchase insurance led to a 60% uptake and access to free insurance to 79% uptake.
Across a range of health measures, there were no significant effects on health.
Clearly, the evidence shows that having health insurance leads to more access to care. But on average this does not lead to better outcomes. How could this be? My theory is that those who are uninsured and very sick get things cared for. I don’t know about other countries, but here, if you show up with a heart attack, stroke, or major trauma, you get care, regardless of insurance.
What is missed when someone is under- or non-insured is preventive care. Perhaps these sorts of studies aren’t detailed enough to show differences in outcomes. Perhaps. But I wonder how much basic health outcomes depend on what we offer for nonemergent care. Perhaps basic health outcomes — not dying early of heart disease or cancer or lung disease — has more to do with societal factors than things like a yearly exam, or vascular or cancer screening. That is, if you live in a community where the norm is to smoke, or not exercise or not go on to learn a trade or get a college degree, it is these factors that lead to higher rates of disease. Maybe it is not access to care. I am not arguing that people should not have access. I am merely suggesting that we consider the possibility that what we offer in non-urgent conditions or asymptomatic people is marginal. The public health implications is that policy and monies might be better spent on societal solutions rather than medical solutions. I like to say build parks not cath labs.
Healthcare journalist and super-nice person Marlene Busko has coverage of a study on testosterone replacement therapy, which was presented at the Endocrine Society meeting in Atlanta and simultaneously published in a Lancet subjournal.
Some quick background: Testosterone exerts all kinds of effects. Some can be beneficial for cardiac healthI — increased coronary blood flow, improved vascular reactivity, increased muscle mass, reduced fat mass, shorter QTc interval, and normalisation of glycemia during lifestyle interventions for prediabetes. But it also can do bad thingsI — increased hematocrit, reduced HDL cholesterol, induction of platelet aggregation, sodium and water retention, and hypertension, as well as smooth muscle proliferation and increased expression of vascular cell adhesion molecules.
Many observational studies have found an association between testosterone supplementation and worse CV outcomes. A small placebo controlled RCT in NEJM 12 years ago found a strikingly higher risk of CV events when testosterone was used in older men who had limited mobility and low testosterone levels. The trial had to be stopped early because for safety reasons. I will link to it in the show notes.
Since testosterone makes you stronger, people like it. The testosterone replacement industry is huge. I know of a number of masters-cyclists busted for testosterone use to improve their watt output. But because of the signals of harm, the US Food and Drug Administration (FDA) has a warning and restricts its approval to hypogonadism caused by documented pituitary or testicular disease, specifically excluding age-related hypogonadism.
The presentation and paper, by a group in Aberdeen Scotland, was a meta-analysis of 35 trials of testosterone vs placebo in men with hypogonadism. While many groups have done meta-analyses of testosterone use, the group in Aberdeen was able to get individual participant data (IPD) in 17 of 35 of the trials. Good on them.
They set out to study CV safety in testosterone trials in men with hypogonadism.
Primary outcomes were all-cause mortality and CV or cerebrovascular events.
Mean age of the 5600 patients was 65 years.
The results were clear: they found no statistically significant differences in CV outcomes; n21o differences in deaths.
So testosterone looks ok, right? Maybe, but the problem here is that the median follow-up was about a year. That is not long enough to make any safety conclusions. This was nicely pointed out in the accompanying editorial from Erin Michos and Matt Budoff.
One of the main things I learned from Marlene Busko’s story and the editorial is that there is a large ongoing CV safety trial called TRAVERSE. This will include more than 500 middle- to old-aged men with low serum testosterone concentrations (<300 ng/dL) who have at least one sign or symptom of hypogonadism and either have established CV risk or are at high risk for CV disease. The primary outcome is a major adverse cardiac event at 60 months. The trial began in 2018 is anticipated to finish in late 2022.
Comments. The TRAVERSE trial may change my mind, but I’ve decided that treating low-testosterone outside of congenital or surgical causes of hypogonadism is a bad idea. This meta-analysis is reasonable. Getting the IPD level data helps. But it’s too short to be reassuring.
Low-testosterone therapy feels like an Ivan Ilyich medical nemesis type social and clinical iatrogenesis. The social iatrogenesis part here is that the pharma-companies, aided by the medical establishment, has medicalized aging. Falling testosterone is part of the natural order. It is probably why older cardiologists are less mean than younger ones. Falling testosterone likely helps keeps marriages together.
The clinical iatrogenesis part is that there are strong signals that giving older men, who are usually overweight, drink too much, and have hypertension, extra testosterone can increase their risk of a serious event—crucially, one that they may not have had if not for modern medicine.
There are natural ways to maintain age-appropriate testosterone levels: stay lean (fat cells are estrogen- making machines), exercise, get good rest, and don’t drink too much alcohol.
Blue Counties and Health
The British Medical Journal has published an observational study looking at the association between how a US county votes and death rate. I realize politics is a hot topic, one I normally stay away from, but I highlight this paper because there are oodles of critical appraisal lessons here. I also want to say right at the outset that I respect the authors. None of my criticisms below should be taken personally.
The authors who have published widely on policy, set out to assess trends in mortality—age adjusted, of course—based on county-level presidential voting patterns.
It was a cross-sectional study done over five presidential elections from 2000 to 2019.
Death rates in Democratic-voting counties fell by 22% over the period while Republican-voting counties only saw an 11% decline in death rates. The gap therefore was substantial.
The greatest contributors to the widening gap between Republican and Democratic counties were heart disease, cancer, and chronic lower respiratory tract diseases.
One peculiar finding — and I think a clue to the limitations — was that Black and Hispanic Americans experienced largely similar improvements in mortality in both Democratic and Republican counties. In fact: “Black residents of Republican counties experienced an improvement in mortality twice that of all other residents in Republican counties.”
The authors were careful not to use overtly causal language in the paper, though it gets close. However, lead author, Haider Warriach, writing in an editorial in Stat, leads with this sentence: “American politics has been creating a deep fissure in the health of Americans over the past two decades.” And this: “In our BMJ analysis, rural Republican counties have the highest death rates and have experienced the least improvement over time. Yet many Republican states have resisted Medicaid expansion.” But I just presented a study that found that Medicaid expansion had no effects on CV outcomes.
Here are the problem with this study:
First, they don’t control for the most glaringly obvious factor in health: socio-economic status. Come on.
Take Kentucky. We are a red Republican state for presidential elections. But there are two small islands of Democratic voting blocks: Louisville and Lexington. The cities. Would it surprise you that these areas, which have much higher average incomes, also have better health outcomes? Wealth is the most obvious factor in health. There are numerous things on the causal path to death rates in a county and how it votes is but one.
Another question. If you are going to posit that Republican policies are deadlier relative to Democratic policies, why were Black Americans not harmed and perhaps better off in Republican counties? This is like a sensitivity analysis that proves confounding.
Again, I mean no personal disrespect to the authors or the BMJ, but this fatally flawed analysis adds nothing to policy debates, and worse, it makes scientists and journals look like activists. That is terrible because, given the massive problem with public trust, now more than ever, medical professionals, scientists and journals ought to strive to be seen as neutral adjudicators of knowledge.
When science comes to be seen partisan, people will become cynical and less apt to believe important findings.
© 2022 WebMD, LLC
Any views expressed above are the author's own and do not necessarily reflect the views of WebMD or Medscape.
Cite this: Jun 10, 2022 This Week in Cardiology Podcast - Medscape - Jun 10, 2022.