David Dobies, MD, FACC, FESC

Interventional cardiologist with an interest in statistics and evidence based medicine. My opinions are not medical advice.

[Ahmed Bennis MD 🫀 @drbennisahmed]

2026 UNIVERSAL DEFINITION Heart Failure

✅ 4 stages (A–D) + recovery/remission pathways

✅ 3 simplified phenotypes: HFrEF, HFpEF, HFimpEF

✅ Universal etiologic classification ischemic, genetic, infiltrative, toxic

✅ AI-enabled ECG & predictive tools for earlier detection

[EHJ-IMP Editor-in-Chief @EHJIMPEiC]

🫀📊 Should we screen asymptomatic patients for coronary artery disease with imaging?
A new systematic review of international guidelines gives a clear (and surprising) answer 👇

✨ Key takeaways:

🚫 No imaging for population screening of asymptomatic CAD
🚫 Not recommended even in high-risk individuals
🧮 Risk scores remain the foundation of primary prevention
🟡 CAC scoring = useful ONLY in selected cases
👉 when risk is uncertain or borderline
🤝 Strong role in shared decision-making with patients
📊 As shown in the graphical summary (page 2):
✅ Agreement → CAC as a risk modifier
⚠️ Disagreement → thresholds & re-testing intervals
❓ Gaps → outcomes, cost-effectiveness, alternative imaging
💡 Big message:
👉 Imaging is not a screening tool
👉 It’s a decision-making tool
🚀 Future direction:
Large trials (e.g. ROBINSCA, CorCal) may finally tell us
if imaging-guided prevention improves outcomes
📄 Read the full paper:
👉
🔗 DOI: 10.1093/ehjimp/qyag034

#Cardiology #Prevention #CAD #CACscore #Guidelines #CardioTwitter #PrimaryPrevention #Imaging #EvidenceBasedMedicine #MedEd 🩺📊

[Anish Koka, MD @anish_koka]

On June 30, 2026, the NEJM retracted the clinical trial that got a drug approved by the FDA, triggering a $3.7B acquisition by Amgen.

The reason: the primary endpoint data was manipulated.

Here's how it happened, how it came to light, and who paid the price. 🧵

[Frank Harrell @f2harrell]

"Real world data" is an almost meaningless phrase. Our new @bmj_latest paper pushes the more helpful phrase "routinely collected data" and explains the many limitations of such for research purposes. https://t.co/VvsxLQvhfk #Statistics

[David Dobies, MD, FACC, FESC @DavidDobiesMD]

@dr_bandak You raise a very valid concern. Statistics needs to be better taught in medical school. The goal should be proficiency. Also professional societies need to teach trial methodology. For example calculate the false positive risk of the p value.

[David Dobies, MD, FACC, FESC @DavidDobiesMD]

@DrJMarine @venkmurthy Once you learn statistics you can quantify uncertainty. This can be very humbling for some people. It is not about how convinced about what you believe.

[David Dobies, MD, FACC, FESC @DavidDobiesMD]

A very informative publication.

[David Colquhoun on Bluesky 🇺🇦 💙 @david_colquhoun]

This is a terrific paper, by four of the best people in the field of statistical inference.
#statistics #bayes #reversebayes
https://t.co/MNqi2bpCpP

[David Dobies, MD, FACC, FESC @DavidDobiesMD]

@pash22 @carole_lunny The guideline committees need to do a systematic review not an informal review. The protocol needs to be published before the guidelines are published.

[David Dobies, MD, FACC, FESC @DavidDobiesMD]

@pash22 @adamcifu A Bayesian approach is very helpful. What is the baseline risk for an individual patient and what are the risks and benefits of the treatment. A Bayesian analysis can help quantify and aid in shared decision making.

[David Dobies, MD, FACC, FESC @DavidDobiesMD]

I published a paper on the limitations of the Heart score. Overlapping confidence intervals. The risk is multiplicative not linear.

[Dorian L. Beasley MD, FACC @cardiojaydoc02]

You can have a HEART score of 2 and ischemic chest discomfort and a HEART score of 4 and have non-ischemic discomfort.

It comes down to the history—the story, the story, the story. #foamed #meded

[Dorian L. Beasley MD, FACC @cardiojaydoc02]

No cardiologist has ever calculated or documented a HEART score.

[David Dobies, MD, FACC, FESC @DavidDobiesMD]

@mirvatalasnag @djc795 @twj1974 @mmamas1973 @nicolasamabile @NievesGonzalo1 @DrNataliaP @SripalBangalore How much cross over to IVUS imaging? The conclusion of the study probably depends upon patient selection.

[David Dobies, MD, FACC, FESC @DavidDobiesMD]

@sbrugaletta Distal LM or mid shaft lesions? Usually distal LM disease involves either the ostial LAD or the ostial LCx which benefits from a POT technique.

[David Dobies, MD, FACC, FESC @DavidDobiesMD]

@kaulcsmc @Bhcintervention From what was posted lesion morphology was not described. If these are mainly non-calcified mid shaft lesions that is not the same as distal LM calcified disease.

[David Dobies, MD, FACC, FESC @DavidDobiesMD]

@DFCapodanno @thiele_holger Lesion morphology was not described in the NEJM abstract presented. Distal LM or mid shaft LM? Calcification?

[Math Files @Math_files]

257 years ago, Joseph Fourier was born on this day.

In 1822, Fourier showed that any wave can be broken down into an infinite sum of sine waves, using a technique now called the Fourier transform.

The Fourier transform is like a recipe generator. You input a complicated wave and you get back its ingredients, the amplitude and frequency of each component sine wave.

Fourier analysis is an essential part of modern technology. Its applications range from JPEG compression and image recognition to quantum physics and MRI's.

Fourier theory has two components: basic building blocks and labels. Imagine a child's toy castle. This castle can be disassembled into individual building blocks. And these building blocks can then be sorted by color into bins and labeled.

Similarly, the Fourier transform disassembles a complex wave into individual sine waves. These are like the building blocks.

Each sine wave can be labeled with its frequency, or how quickly it oscillates per second. The labels on the bins are more than just a way to organize things.

They can be used to rebuild the original complex wave and as an efficient shorthand for communicating information.

For example, when you send a voice message, your phone doesn't transmit an entire complex sound wave. Instead, it breaks it down and sends just the labels or frequencies of the component sine waves.

The receiver's phone then reverses this process, converting the labels back into the contents of the bins to reconstruct the message’s original sound wave.

[Frank Harrell @f2harrell]

Fantastic new paper casting doubt on explainability of explainable AI: https://t.co/WSfMC6NjNp @MaartenvSmeden #machinelearning #Statistics

[Venk Murthy MD PhD @venkmurthy]

These guidelines are simply grade inflation

[Edward Hulten @cardiac_md]

2012 USPSTF: insufficient evidence for CAC

2013 ACC/AHA: Class IIb rec for CAC

2018 USPSTF: insufficient evidence for CAC

2018 ACC/AHA: Class IIa rec for CAC

2019 JAMA review: insufficient evidence for CAC commonly ordered for executive physicals.

2023 ACC/AHA class IIa rec for cac

- With no new clinical trial or any compelling evidentiary change,

2026 ACC/AHA: FIVE Class I rec for CAC.

Huh? 🤷‍♂️

https://t.co/nascLb3iyZ

[David Dobies, MD, FACC, FESC @DavidDobiesMD]

@LocasaleLab Why not require data sharing in order to publish?