Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
The report is confusing. As I read it, only 34% of people who got irregular heart rhythm warning from AW actually had atrial fibrillation. That's 66% false positive rate which is unacceptable.

The study was overall beneficial, demonstrating the potential for large-scale studies that use a variety of technologies to monitor patients remotely without requiring on-site visits. Since this study kicked off in 2017, it did not use the new Apple Watch models that are able to take ECG readings, instead relying on the standard heart rate sensor.

Seems like a win. Of course, one doesn't have to get an apple watch or participate in the studies. Perfectly acceptable. But for those who do, it just may save their life if they are an at-risk patient.
 
  • Like
Reactions: Solomani
Having recently found I have an irregular heartbeat of 1250 in 100,000 it'd be great if this passed medical scrutiny in Australia and I'd be waiting in line for version 6 of the watch.
I agree, I hope it’s not too long before the ECG function on the Apple Watch is approved here, but the report notes it commenced before ECG was added to the watch, and relied only on the watch’s heartbeat monitor.
 
That's 66% false positive rate which is unacceptable.

As a person who had atrial fibrillation misdiagnosed by 7 heart specialists over a 12 year period I know how random the attacks can be. So when you say "66% false positive rate" I would argue that much of that is a false reading by review specialists as the an "attack" rytham was not happening at time of review.
 
Why is it unacceptable? Based on what comparison? It’s a screening evaluation where the downsides are minimal and the upside is huge - tests typically designed to cast a wider net than other evaluations. They also say some may have had atrial fibrillation but it was so rare the watch caught it but the other testing didn’t.

As a doctor, understanding false positive rates and how to use screening tests is important. A lot of patients ask me "Why don't you just scan my whole body doc?" or test me for this or that.

The harms of testing are often underestimated by patients. Whenever a screening test is positive, physicians have to think about the anxiety of the patient, the time the patient must take to address this problem (take off work etc), and then the additional harms caused by further testing to confirm a positive screening test (which are inherently becomes more invasive).

I think the study showed a fair positive predictive value but we need to see if it actually improves a patient oriented outcome. Screening likely needs to be age oriented.

Why don't we screen all women at all ages for breast cancer? We don't because we balance the benefits and the harms and have decided that a 25 year old would have more harms and unnecessary follow-up testing versus 50 year old women. Screening can be inappropriate for different populations.

I bet this will happen with atrial fibrillation. If we prove that there is a potential benefit of identifying atrial fibrillation early (prevention of strokes or mortality improvement) and then specifying who would receive the benefit the most (likely people >65 years old), I could see the Apple watch as being useful in health screenings.
 
Atrial fibrillation (A-fib) is only one example of an irregular rhythm (technically it's "irregularly irregular" but anyway), and having a brief arrhythmia is normal in the general population. The follow-up ECG patches were received by participants an average of thirteen days after the original watch notification alerted the research team.

So two weeks after the original watch notification, the ECG patches recorded a diagnosable rhythm in 34% of the 450 patients who returned the patches. While those 34% of the 450 had the ECG patches applied, irregular heart rhythm notifications received on the watch correlated with A-fib on the "real-time" ECG 84% of the time.

If the irregular notification "accuracy" is 84%, we could speculate that the 50% of people who did not show A-fib on the follow-up ECG patches, did in fact have transient A-fib that resolved and did not present again. That would not be unusual, however there's no way to know definitively since two weeks passed between notification and ECG collection.


Sorry that you posted before I finished typing, but using your simplified example: 100 received an alert, 34 had confirmed A-fib, and of those 34, 29 had A-fib alerts on their watch that matched A-fib on the ECG.
The other 66 have an unknown status, and we could guess that 50 of them had A-fib that would have shown up on the ECG, had they been wearing one at the time of the alert.

Clear as mud?? 😆
Thank you, that helped a lot.
 
  • Like
Reactions: Neberheim
I agree, I hope it’s not too long before the ECG function on the Apple Watch is approved here, but the report notes it commenced before ECG was added to the watch, and relied only on the watch’s heartbeat monitor.

You may be waiting a very long time. Latest news from September, shortly after the Series 5 launch, is Apple hasn’t even submitted the ECG functionality to the TGA in Australia for approval.
 
I guess if people were to die, you'd more likely tell the truth, so ya ok,, i beleive this study :)
 
This is actual good stuff! The results are just OK, but look at the number of participants and the timeframe. They were able to recruit 420K people over 8 months. That's 1700 people a day! This is a proof of concept study showing how high quality wearables can speed up clinical studies. This is the beginning of a new era for public health research.
[automerge]1573705595[/automerge]
As a doctor, understanding false positive rates and how to use screening tests is important. A lot of patients ask me "Why don't you just scan my whole body doc?" or test me for this or that.

The harms of testing are often underestimated by patients. Whenever a screening test is positive, physicians have to think about the anxiety of the patient, the time the patient must take to address this problem (take off work etc), and then the additional harms caused by further testing to confirm a positive screening test (which are inherently becomes more invasive).

I think the study showed a fair positive predictive value but we need to see if it actually improves a patient oriented outcome. Screening likely needs to be age oriented.

Why don't we screen all women at all ages for breast cancer? We don't because we balance the benefits and the harms and have decided that a 25 year old would have more harms and unnecessary follow-up testing versus 50 year old women. Screening can be inappropriate for different populations.

I bet this will happen with atrial fibrillation. If we prove that there is a potential benefit of identifying atrial fibrillation early (prevention of strokes or mortality improvement) and then specifying who would receive the benefit the most (likely people >65 years old), I could see the Apple watch as being useful in health screenings.

But this is so much more than screening. This is a realtime longitudinal monitoring tool, that can potentially provide more clinically relevant data than a cross sectional screening tool. Just think about the results of CRYSTAL AF...It showed in a cryptogenic stroke population that longer monitoring allows higher detection rate of atrial fibrillation. Yet the stroke risk with paroxysmal AFib is not that clear in the general population. Questions like this could be answered using a similar design, within a year or two. It's insanely useful, even with a marginally good device.
 
Last edited:
  • Like
Reactions: bobob
Wow! I am impressed. Damn good job, Apple! The watch is only going to get better. I can't wait to see what it does in 5 years or so.
Call me vain, but if the watch looked less like a large mint-lolly I'd buy one tomorrow.
 
You may be waiting a very long time. Latest news from September, shortly after the Series 5 launch, is Apple hasn’t even submitted the ECG functionality to the TGA in Australia for approval.
That’s what they said about Canada up until the week before it was approved, so there’s still hope :)
 
  • Like
Reactions: matrix07
Dr. Daniel Cantillon, a Cleveland cardiologist who was not involved, told Reuters that the technology was promising, but more than half of participants were under 40, a group at low risk for atrial fibrillation, leading to concerns about scaring healthy people.

Separately, a New York cardiologist told CNBC that there's a risk of the Apple Watch finding young people who have early signs of atrial fibrillation that the medical community doesn't know how to treat. "We just don't understand atrial fibrillation well in the 35-year-old, otherwise healthy person," he said.

These two items are very important to take into account - as someone that has some cardiological issues, the healthcare INDUSTRY like to promote various fixes. Just as the rise in orthorexia falling lock-step with the food INDUSTRY, the reliance on electronic devices for monitoring our health appears to be well-intentioned, but can have significant unwanted side-effects.

Although I am far from a Luddite (I literally have a satellite that I worked on in space), I am weary of this type of technology causing undue stress when we don't actually know what to do with the results or even what the numbers really mean*

*using the cholesterol levels as an example - I remember when all cholesterol was bad - then only some, then the ratios, and now perhaps none of it matters w.r.t all-cause mortality. In the mean time we cut out eggs, avocados etc...

Lack of an association or an inverse association between low-density-lipoprotein cholesterol and mortality in the elderly: a systematic review

 
Is the New England Journal of Medicine part of Apple News or is everyone basing their opinions on the summary? /s

Stupid paywall.
 
You may be waiting a very long time. Latest news from September, shortly after the Series 5 launch, is Apple hasn’t even submitted the ECG functionality to the TGA in Australia for approval.

The lives of some customers are more important than others to Apple.

Also Apple didn't even start any submissions until after the keynote. Having it be a surprise feature was more important than the lives of people who could have benefited from the day it was released.
 
I was in the study, and have to pay for the report? Harsh...
Non-open access research papers are a problem in the scientific community. However, you can always email the research and kindly ask for a PDF copy of the paper (maybe congratulating them on the great work while you're at it). Researchers are more than happy to do this and the publishers (here NEJM) usually does allow this kind of individual peer-to-peer sharing.
 
You're so vain, you probably think this watch is about you.
Yep, I do. Well, actually I think the Garmin is about me more, but that's beside the point. I think it's good however that Apple's bio monitoring is making real advances.
 
Half baked journalism by MR here. Getting the Apple Watch (the software, technically) cleared by FDA under a de novo approval for AFib detection already required Apple to prove safety and efficacy of the device for its therapeutic claim. It means that they already demonstrated that “it works”. This study was more of a double check and not really that groundbreaking of news.

FDA’s job is literally to police companies making therapeutic claims to ensure that the claims are not baseless or inflated.
 
The report is confusing. As I read it, only 34% of people who got irregular heart rhythm warning from AW actually had atrial fibrillation. That's 66% false positive rate which is unacceptable.

I had AF for 5 years. Was diagnosed when I was 33.
The problem I had was that when I got the equipment to monitor my heart for 24 hours, my heart was always in rhythm.
It wasn’t until once it went out of rhythm and I went to emergency where they recorded it so I could be properly diagnosed.
I had ablation surgery to kill the rouge cells in my heart in 2012, and now I am completely cured.

So there is a chance the ‘false positives‘ are actually real events that have not developed into a significant problem yet.
 
The lives of some customers are more important than others to Apple.

Also Apple didn't even start any submissions until after the keynote. Having it be a surprise feature was more important than the lives of people who could have benefited from the day it was released.
There may be reasons beyond our own biases, that Apple can’t (or didn’t want to)actually submit paperwork in every Country simultaneously for reasons of conservatism.

But the naysayers spin that into some sort of negativity. however Apple is steady as she goes, so carry on.
 
I was one of the false positives who wore and actually returned the ECG patch. The doctor I spoke to post-ECG said that they were very happy to have the data, because I am a distance runner. My arrhythmia warning came overnight, when my HR dips into the 40's, sometimes the low 40's. She said it might help them reduce false positives.

(The patch was a bit annoying to wear, so I'm not too surprised that compliance was low. You had to shave a patch on your chest, and apply the patch. It itched as the hair tried to grow back. And after a few days it started to come loose, so I had to reapply the device with the second patch they included.)
 
  • Like
Reactions: Khedron
Non-open access research papers are a problem in the scientific community. However, you can always email the research and kindly ask for a PDF copy of the paper (maybe congratulating them on the great work while you're at it). Researchers are more than happy to do this and the publishers (here NEJM) usually does allow this kind of individual peer-to-peer sharing.

I tried that once. I emailed a local state uni researcher, and he said to email the primary researcher, and they said to email their office of research, and they said to email the second primary researcher, and I got nowhere.

Yes, our tax dollars pay for so much research, and the idea that we then have to pay again for access to the research is so obnoxious. Pay twice, why not... They make a lot of money off subscriptions, and access from what I've heard. Corporations after the research budget that cost but you and me often can't afford the cost. There have been attempts to regulate the industry, but it always dies before anything changes for the better.
 
My background: somewhat athletic 58-year-old PhD in biophysics with past episodes of AF in 2015-2017, using a competing device (Kardia) for routine EKG self-monitoring.

Separately, a New York cardiologist told CNBC that there's a risk of the Apple Watch finding young people who have early signs of atrial fibrillation that the medical community doesn't know how to treat. "We just don't understand atrial fibrillation well in the 35-year-old, otherwise healthy person," he said.
I'll second that. We have epidemiological data showing that middle-aged endurance athletes like commenter JHFenton above (and even younger ones like RandyJ) are at increased risk of AF, even though all their other cardiac risks are lower thanks to exercise. We don't have a good understanding of the mechanism for it though. It may have some connection with the relatively low resting heart rate that many of these people have.

This is where putting a simple single-lead heart rhythm measurement tool in (or on!) the hands of patients has a lot of value. Patients don't always recognize the symptoms of persistent AF, or ascribe them to other things, and it's only after several days that they see their doctor and get diagnosed. That's enough time for a blood clot to develop in the irregularly-beating heart and lead to a stroke or other dangerous complication. The watch really will save lives.

If the irregular notification "accuracy" is 84%, we could speculate that the 50% of people who did not show A-fib on the follow-up ECG patches, did in fact have transient A-fib that resolved and did not present again. That would not be unusual, however there's no way to know definitively since two weeks passed between notification and ECG collection.

Another good point. Many episodes of AF resolve spontaneously in a matter of minutes or hours, so we wouldn't expect the follow-up monitoring (7 day continuous recording) to be 100% positive, even if the watch worked perfectly.

This points out another benefit of the wearable device: patients who have had a past AF episode can take an ECG whenever something feels odd or they're having heart-related symptoms. They can then see whether there's another AF episode going on, some non-persistent irregularity in the ECG, or an unusually low heart rate with normal rhythm (sinus bradycardia). In my case, my cardiologist and I have a management plan where I don't have to take any medication unless I see a persistent AF episode.

As a doctor, understanding false positive rates and how to use screening tests is important. A lot of patients ask me "Why don't you just scan my whole body doc?" or test me for this or that.

Everyone read that carefully too! Indiscriminate use of screening tests does bring on small but real harms, such as the risk of complications from a biopsy, or side effects from treating a cancer that is so small and growing so slowly that the patient will die of other causes before the cancer becomes clinically detectable.
 
  • Like
Reactions: jhfenton
Two weeks ago I noticed a bout of flutters in my heart that occasionally happen, but the over the last two days they became worse and for longer and more frequent periods. I recorded several episodes on the Apple Watch 4, and visited the military ER. The ER ran EKGs which showed signs of Afib. When talking to the doctor I stated I had some recordings from the Watch and iPhone. He was blown away at the look of the ECG in comparison to the EKG data and had several colleagues come in to take a look. The EKG, and expertise of the emergency staff and doctors initiated my followup with a cardiologist next week, but without a doubt the Watch and it's data scared me enough to have everything checked out.
 
  • Like
Reactions: citysnaps
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.