Feb 12, 2021 This Week in Cardiology Podcast

John M. Mandrola, MD


February 12, 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 February 12, 2021, John Mandrola, MD comments on the following news and features stories.


Before I get to the study published in JAMA-Internal Medicine, let’s set out that hospitals and clinicians have enough manifest diseases to treat without creating extra categories for those at risk for a condition, like prediabetes or pre-hypertension. Precondition states are nuts. Not dying at a young age puts you are risk of heart failure, along with hypertension and diabetes and cancer and falls and dementia and COVID. Now to the timely and worthy study:

Dr. Mary Rooney from Johns Hopkins, and colleagues, used data from the ARIC longitudinal study to look into risk of progression from prediabetes to diabetes in older adults. ARIC is one of those population-based studies (like Framingham) that follow patients over time. This study included about 3400 older aged adults (71-90 years) without diabetes. The authors defined prediabetes: by HbA1c, impaired fasting glucose, or both. Individuals were followed for 6.5 years and the primary endpoint was developing overt diabetes defined again by HbA1c, or glucose-lowering medicine, or fasting glucose.


  • The prevalence of prediabetes ranged from roughly 30% to 65% depending on the definition used. Bottom-line: there was a lot of prediabetes.

  • Among participants with prediabetes HbA1c levels, only 9% progressed to diabetes, 13% regressed to normoglycemia, and 19% died.

  • Similar patterns were seen in the other categories of prediabetes.

Thus, older adults in this fairly representative cohort had a small chance of progressing to true diabetes. In fact, there was a higher risk of regressing to normal than progressing, and a more than double chance of dying than progressing to diabetes. The authors humbly conclude that prediabetes may not be a robust diagnostic entity in older age.

The specific translation of this study, and others like it, are to recognize that for older adults with limited life expectancy, prediabetes is irrelevant; consider placing less emphasis on these numbers in healthier older adults; and revisit the value of prediabetes in middle-aged adults.

Mobile QT interval Measurement

The Mayo Clinic group is doing incredible work in applying artificial intelligence (AI) to electrocardiogram (ECG) interpretation–not only on standard 12-lead ECGs, but also smartphone-enabled ECGs.

Before I tell you about the paper, two background items are notable.

  • While we mostly reserve implantable cardioverter defibrillators (ICDs) for patients with low ejection fractions (EF), the vast majority of sudden cardiac death, in absolute terms, occurs in people with EFs better than 30%-35%. That’s because there are far more people with good EFs than bad EFs.

  • We know that virtually anything (except amiodarone and V-pacing) that pushes the QTc above 500 ms (genetics, drugs, hypokalemia, cancer, and even COVID-19) is a pro-mortality Indicator.

The Mayo group used about 1.6 million standard 12 lead ECGs from more than a half-million patients seen there to derive a deep neural network (DNN). They compared the DNN system to cardiologists’ over-read of QTc as the gold standard.

The ability of this DNN to detect clinically relevant QTc prolongation (eg. QTc ≥ 500 ms) was then tested prospectively on about 700 patients with genetic heart disease (GHD; half with long QT syndrome) with QTc values obtained from both a 12-lead ECG and a prototype mobile ECG (mECG) device equivalent to the commercially-available AliveCor KardiaMobile 6L.

  • In the validation sample, they saw strong agreement between human over-read and DNN-predicted QTc values; the authors called these nominal differences.

  • In the group of patients seen in the genetic clinic, the DNN-predicted that QTc values from the mobile ECG were similar to those read from a 12-lead by a QT expert and commercial core ECG lab.

  • When applied to mECG tracings, the DNN’s ability to detect a QTc value ≥ 500 ms yielded an area under the curve, sensitivity, and specificity of 0.97, 80.0%, and 94.4%, respectively.

You should read both the paper and Steve Stiles’ coverage. His is a great piece; he gets relevant comments from both the senior author (an enthusiast) and a neutral observer.

In the paper they propose using mECG for futuristic things like spot checks in pharmacies when a patient is given a strong QT prolonging drug. One obvious limitation of this technology would be in the realm of screening low-risk populations. Keep in mind, they tested it in a population enriched with long QT syndrome patients.

Enter Bayes Theorem. Recall that medical tests don’t tell you if you have a disease, it just updates your prior probability. By my calculation, in a population with a 1% incidence of long QT syndrome, the DNN-enabled mECG would update your 1% probability of having a long QT by 14, to about 0.14 odds or 12% probability. Thus, more than 85 out of 100 positives would be false positives.

I may be wrong, but I have a similar feeling about this whole AI in ECGs thing as my initial thoughts on cardiac resynchronization therapy when someone first asked me, many many years ago, whether putting a lead in the coronary sinus could reverse left bundle branch block and improve heart failure. I said wrongly, that would never work.

Thus, being too pessimistic here might be the wrong take.

Heart Failure With Preserved Ejection Fraction (HFpEF)

Yet another RCT in patients with HFpEF has been published. The investigators set the stage for a good outcome: they screened more than 500 patients and enrolled 180 presumably motivated patients. They measured a highly relevant primary outcome, peak VO2. I like peak VO2 because it quantifies exercise tolerance. They enrolled representative patients; 70% were women, age about 71 years, with a body mass index (BMI) of 30, and a good mix of co-morbid conditions.

Yet they found not even a tiny signal of benefit. Despite telemedical support, only about one-half of the patients performed at least 70% of the prescribed training sessions during home-based exercise training (months 4-12). But this is part of the deal with exercise or diet interventions–part of it is the physiologic effects and the other part is the adherence. If you had a wonder drug that delivered huge benefits, but 30% of people couldn’t take it because it burned the esophagus, it would be a lousy pill. Walking and regular exercise is always a wise prescription, but I still hold out hope for the SGLT2 inhibitors in this group of patients.

Observational Study Follies

A group of well-meaning authors looked back retrospectively, using two datasets, at patients who had adhesion-related abdominal issues. They found that statin use was associated with a 19% reduction in one data set and 8% reduction in the other in adhesion related issues in the belly. The authors did all the usual corrections and adjustments.

The authors concluded, “Statins may represent an inexpensive, well-tolerated pharmacologic option for preventing ARCs (adhesion related complications).” This is awful. You simply cannot make these sorts of highly improbable causal claims—that statins reduce adhesions—without a proper randomized trial. I cannot understand is how referees and editors let the authors use such causal language.

There are very good reasons to take a statin, but reduction of adhesion issues is not one of them.


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.
Post as: