The skeptical cardiologist primarily makes decisions on blood pressure treatment these days based on patient self-monitoring. If high readings are obtained in the office I instruct patients to use an automatic BP cuff at home and make a measurement when they first get up and again 12 hours later. After two weeks they report the values to me.
Although I’ve been recommending self-monitoring to my patients for decades it is only recently that guidelines have endorsed the approach and good scientific studies verified its superiority. I was pleased when the 2017 ACC/AHA guidelines for High Blood Pressure made home self-monitoring of BP a IA recommendation.
And last year a very good study, the TASMNH4 was published which demonstrated the superiority of self-monitoring compared to usual care.
TASMINH4 was a parallel randomised controlled trial done in 142 general practices in the UK, and included hypertensive patients older than 35 years, with blood pressure higher than 140/90 mm Hg, who were willing to self-monitor their blood pressure. Patients were randomly assigned (1:1:1) to self-monitoring blood pressure (self-montoring group), to self-monitoring blood pressure with telemonitoring (telemonitoring group), or to usual care (clinic blood pressure; usual care group).
The home BP goal was 135/85 mm Hg, 5 mm Hg lower than the office BP goal. At one year both home self-monitoring groups had significantly lower systolic blood pressure than the usual care group.
This trial was not powered to detect cardiovascular outcomes, but the differences between the interventions and control in systolic blood pressure would be expected to result in around a 20% reduction in stroke risk and 10% reduction in coronary heart disease risk. Although not significantly different from each other at 12 months, blood pressure in the group using telemonitoring for medication titration became lower more quickly (at 6 months) than those self-monitoring alone, an effect which is likely to further reduce cardiovascular events and might improve longer term control.
Advantages of Home Self-Monitored Blood Pressure-Limitations of Office BPs
Every patient I see in my office gets a BP check. This is typically done by one of the office assistants who is “rooming” the patient using the classic method with , listening with stethoscope for Korotkov sounds. If the BP seems unexpectedly high or low I will recheck it myself.
Often the BP we record is significantly higher than what the patient has been getting at home or at other physician offices.
There are multiple factors that could be raising the office BP: mental stress from driving to the doctor or being hurried or physical stress from walking from the parking lot.
In addition, I feel that multiple assessments of out of office BP over the course of the day and different days are more likely representative of the BP that we are consistently exposed to rather than one reading in the doctor’s office.
Accuracy and technique in the doctor’s office is also an issue.
Interestingly, we have assumed that manual office BP measurement is superior to automatic but this recent paper found the opposite:
Automated office blood pressure readings, only when recorded properly with the patient sitting alone in a quiet place, are more accurate than office BP readings in routine clinical practice and are similar to awake ambulatory BP readings, with mean AOBP being devoid of any white coat effect.
A patient left a comment to that paper which is quite insightful:
I had a high blood pressure event several years ago. Since then I have monitored my BP at home, sitting with both feet flat on the floor, not eating or drinking, not speaking or moving around, on a chair with a back, and without clothes on the arm being used for the measure. My BP remains normal.
I have never had my BP taken correctly in a doctor’s office. They will do it while I am speaking with the doctor, sitting on an exam table with my legs swinging, with the monitor band over my heavy winter sweater, right after I have sat down. They do not ensure that my arm is supported or at the right height. If I recommend that I take off my sweater, or move to a chair with a back, they tell me that is not needed. I have decided to refuse such measurements. How can they possibly be monitoring my health this way?
This patient’s observations are not unique and I suspect the majority of office BPs have most if not all of the limitations she describes.
Self Monitoring Improves Patient Engagement In BP Control
I have found self-monitoring of patient’s BP to substantially enhance patient engagement in the process. Self-monitoring patients are more empowered to understand the lifestyle factors which influence their BP and make positive changes.
Blood pressures are amazingly dynamic and as patient’s gain understanding of what influences their BP they are going to be able to take control of it.
I take my BP almost daily and adjust my BP medications based on the readings. After prolonged work or exercise in heat, for example, BPs will decline to a point where I’m light headed or fatigued. Less BP med at this time is indicated. Conversely, if I’ve been overly stressed BPs increase and upward titration of medication is warranted.
With some of my most engaged and enlightened patients we perform similar titrations depending on their circumstances. Sometimes patients perform these titrations on their own and tell me about them at the next office visit.
What’s The Best Way To Communicate Home BPs?
Many of my patients provide me with a hand-written record of their BPs over two weeks. Some mail them to me, others bring them in to the office. We scan these into the EMR. I look at these and make an estimate of the average systolic blood pressure, the variation over time and the variation during the day. It’s not feasible for me or my staff to enter the numbers or precisely obtain an average.
Some patients send us the numbers through the internet-based patient portal into the EMR. This is preferable as I can view these and respond quickly and directly back to the patient with recommendations.
More and more patients are utilizing their smart phones to record and aggregate their health data and will bring them in for me to look at during an office visit. I’ve described one stylish and slick BP cuff, the QardioArm which has neither tubes nor wires and works through a smartphone app. Omron , also has multiple cuffs which communicate via BlueTooth to store data in a smartphone app.
Ideally, we would have a way for me to view those digitally recorded BPs with nicely calculated averages online and within the EMR. Unfortunately, such connectivity is not routinely available.
However, for my patients who are already monitoring their heart rhythms with a Kardia mobile ECG and are connected with me online through KardiaPro Remote I can view their BP recordings online.
I’ll discuss in detail in a subsequent post the Omron Evolv home automatic BP cuff (my current favorite) which is wireless and tubeless and connects seamlessly to KardiaPro allowing me to view both BP and heart rhythm (and weight) recordings in my patients
To me, this empowerment of patients to record, monitor and respond to their own physiologic parameters is the future of medicine.
From the 2017 ACC/AHA BP guidelines
and the proper technique for office BP measurement
In less than a month AliveCor plans to release its KardiaMobile 6L which will provide 6 ECG leads using a smartphone based mobile ECG system that is similar to the Kardia single lead system.
AliveCor’s website proclaims “This is your heart x 6.”
I was fortunate enough to obtain a demo version of the 6L and have been evaluating it.
My first impressions are that this is a remarkable step forward in the technology of personal ECG monitoring. I’m not sure if I would call it “your heart x 6” but I feel the ability to view six high quality leads compared to one is definitely going to add to the diagnostic capabilities of the Kardia device.
Kardia 6L Setup And Hardware
The 6L is similar in design and function to the single lead device.
I’m including this cool spinning video (from the AliveCor website) which makes it appear, slick, stylish and futuristic
Once paired to the Kardia smartphone app (available for iOS or Android smartphones for free) it communicates with the smartphone using BLE to create ECG tracings.
Like the single lead Kardia the 6L has two sensors on top for left and right hand contact. But in addition, there is a third on the bottom which can be put on a left knee or ankle.
The combination of these sensors and contact points yield the 6 classic frontal leads of a full 12 lead ECG: leads I, II, III, aVL, aVR, and aVF. This is accomplished, AliveCor points out “without messy gels and wires.
I found that using the device was simple and strait-forward and we were able to get high quality tracings with minimal difficulty within a minute of starting the process in all the patients we tried it on.
The Diagnostic Power Of Six Leads
Below is a tracing on a patient with known atrial fibrillation. The algorithm correctly diagnoses it. With 6 different views of the electrical activity of the atrium I (and the Kardia algorithm) have a better chance of determining if p waves are present, thereby presumably increasing the accuracy of rhythm determination
Depending on the electrical vector of the left and right atria, the best lead to visualize p waves varies from patient to patient, thus having 6 to choose from should improve our ability to differentiate sinus rhythm from afib.
In the example below, the Kardia 6L very accurately registered the left axis deviation and left anterior fascicular block that we also noted on this patient’s 12 lead ECG. This 6L capability, determining the axis of the heart in the frontal plane, will further add to the useful information Kardia provides.
For a good summary of axis determination and what abnormal axes tells us see here.
The History of ECG Leads
When I began my cardiology training the 12-lead ECG was standard but it has not always been that way. I took this timeline figure from a nice review of the history of the ECG
Einthoven’s first 3 lead EKG in 1901 was enormous.
It is mind-boggling to consider that we can now record 6 ECG leads with a smartphone and a device the size of a stick of gum
For the first 30 years of the ECG era cardiologists only had 3 ECG leads to provide information on cardiac pathology. Here’s a figure from a state of the art paper in 1924 on “coronary thrombosis” (which we now term a myocardial infarction) showing changes diagnostic of an “attack” and subsequent atrial fibrillation
In the 1930s the 6 precordial leads were developed providing more information on electrical activity in the horizontal axis of the heart. The development of the augmented leads (aVr, aVL, aVF) in 1942 filled in the gaps of the frontal plane and the combination of all of these 12 leads was sanctified by the AHA in 1954.
I’ll write a more detailed analysis of the Kardia 6L after spending more time using it in patient care.
Specifically I’ll be analyzing (and looking for published data relative to):
-the relative accuracy of the 6L versus the single lead Kardia for afib determination (which, at this point would be the major reason for current Kardia users to upgrade.)
-the utility of the 6L for determination of cardiac axis and electrical intervals in comparison to the standard 12 lead ECG, especially in patients on anti-arrhythmic drugs
For now, this latest output from the meticulous and thoughtful folks at AliveCor has knocked my socks off!
N.B. If one uses the single lead kardia device in the traditional manner (left hand and right hand on the sensors) one is recording ECG lead I. However, if you put your right hand on the right sensor and touch the left sensor to your left leg you are now recording ECG lead II and if to the right leg, ECG lead III.
I describe this in detail here. For certain individuals the lead II recordings are much better than lead I and reduce the prevalence of “unclassified” recordings.
My feeling is that by automatically including the leg (and leads II and III) the 6L will intrinsically provide high voltage leads for review and analysis, thereby improving the ability to accurately classify rhythm.
And (totally unrelated to the 6L discussion) one can also record precordial ECG leads by putting the device on the chest thus theoretically completing the full 12 leads of the standard ECG.
Please also note that I have no financial or consulting ties to AliveCor. I’m just a big fan of their products.
Many patients (and perhaps physicians) are confused as to how best to utilize personal ECG devices. I received this question illustrating such confusion from a reader recently:
I first came across your website a year ago during persistent angina attacks, and returning now due to increasing episodes of symptoms akin to Afib. I bought a Kardia 2 yrs ago for the angina episodes, and looking to buy the Apple Series 4 for the Afib, as I want to try a wearable for more constant monitoring. What I would greatly appreciate if you had a basic guide for both the Kardia & Apple devices, specifically when and how to best employ them for unstable angina and detecting undiagnosed Afib. As in, what can I as a patient provide to you as a doctor for diagnosis in advance of a formal visit. I’m a US Iraq vet medically retired in the UK, and most of my concerns get dismissed out of hand as “anxiety”, not sure why they thought a stent would cure my anxiety though
First. please understand that none of these devices have any significant role in the management of angina. Angina, which is chest/arm/jaw discomfort due to a poor blood supply to the heart muscle cannot be reliably diagnosed by the single lead ECG recording provided by the Apple Watch, the Kardia Band or the Kardia mobile ECG device. Even a medical-grade 12 lead ECG doesn’t reliably diagnose angina and we rely on a constellation of factors from the patient’s history to advanced testing to determine how best to manage and diagnose angina.
Second, as you are having episodes “akin to Afib”, all of these devices can be helpful in determining what your cardiac rhythm is at the time of the episodes if they last long enough for you to make an ECG recording.
The single lead ECG recording you can make from the Apple Watch, the Kardia Band and from the Kardia mobile device can very reliably tell us what the cardiac rhythm was when you were feeling symptoms.
The algorithms of these devices do a good job of determining if the rhythm Is atrial fibrillation. Also, if the rhythm is totally normal they are good at determining normality.
These tracings can be reviewed by a competent cardiologist to sort out what the rhythm really is.
In all of these cases, having an actual recording of the cardiac rhythm at the time of symptoms is immensely helpful to your doctor or cardiologist in determining what is causing your problems.
My recommendation, therefore, would be to make several recordings at the time of your symptoms. Print them out and carefully label the print-out with exactly what you were feeling when it was recorded and present these to the doctor who will be reviewing your case.
As I’ve mentioned in previous posts (see here), my patients’ use of Kardia with the KardiaPro online service has in many cases taken the place of expensive and inconvenient long term monitoring devices.
Case Example-Diagnosing Rare And Brief Attacks Of Atrial Fibrillation
I recently saw a patient who I think perfectly demonstrates how useful these devices can be for clarifying what is causing intermittent episodes of palpitations-irregular, pounding, or racing heart beats.
She was lying on a sofa one day when she suddenly noted her heart “pumping fast” and with irregularity. The symptoms last for about an hour. She had noticed this occurred about once a year occurring out of the blue.
Her PCP ordered a long term monitor, a stress test and an echocardiogram.
The monitor showed some brief episodes of what I would term atrial tachycardia but not atrial fibrillation but the patient did not experience one of her once per year hour long episodes of racing heart during the recording. Thus, we had not yet solved the mystery of the prolonged bouts of racing heart.
She was referred to me for evaluation and I recommended she purchase an Alivecor device and sign up for the KardiaPro service which allows me to view all of her recordings online. The combination of the device plus one year of the KardiaPro service costs $120.
She purchased the device and made some occasional recordings when she felt fine and we documented that these were identified as normal by Kardia. For months nothing else happened.
Then one day in April she had her typical prolonged symptom of a racing heart and she made the recording below (She was actually away from home but had the Kardia device with her.)
When she called the office I logged into my KardiaPro account and pulled up her recordings and lo and behold the Kardia device was correct and she was in atrial fibrillation at a rate of 113 BPM.
With the puzzle of her palpitations solved we could now address proper treatment.
Continuous Monitoring for Abnormal Rhythms
Finally, let’s discuss the wearables ability to serve as a monitor and alert a patient when they are in an abnormal rhythm but free of any symptoms.
My reader’s intent was to acquire a device for “constant monitoring”:
I’m looking to buy the Apple Series 4 for the Afib, as I want to try a wearable for more constant monitoring.
This capability is theoretically available with Apple Watch 4’s ECG and with the Kardia Band (using SmartRhythm) which works with Apple Watch Series 1-3.
However, I have not been impressed with Apple Watch’s accuracy in this area (see here and here) and would not at this point rely solely on any device to reliably alert patients to silent or asymptomatic atrial fibrillation.
In theory, all wearables that track heart rate and alert the wearer if the resting heart rates goes above 100 BPM have the capability of detecting atrial fibrillation. If you receive an alert of high HR from a non ECG-capable wearable you can then record an ECG with the Kardia mobile ECG to see if it really is atrial fibrillation.
At 99$, the Kardia is the most cost-effective way of confirming atrial fibrillation for consumers.
I hope this post adds some clarity to the often confusing field of personal and wearable ECG devices.
The skeptical cardiologist recently received this email from a reader:
With the new Apple Watch that’s out now, people have suggested my husband (who had a heart attack at 36) should get it since it could detect a heart attack. But I keep remembering what you said – that these devices can’t detect heart attacks and that Afib isn’t related to a heart attack most of the time – is that still the case? I don’t really know how to explain to people that it can’t do this, since absolutely everyone believes it does.
The answer is a resounding and unequivocal NO!
If we are using the term heart attack to mean what doctors call a myocardial infarction (MI) there should be no expectation that any wearable or consumer ECG product can reliably diagnose a heart attack.
The Apple Watch even in its latest incarnation and with the ECG feature and with rhythm monitoring activated is incapable of detecting a myocardial infarction.
To make this even clearer note that when you record an ECG on the Apple Watch it intermittently flashes the following warning:
Note: “Apple Watch never checks for heart attacks”
How did such this idea take root in the consciousness of so many Americans?
Perhaps this article in 9-5 Mac had something to do with it
In reality, the man received an alarm that his resting heart rate was high at night. Apparently he also was experiencing chest pain and went to an ER where a cardiac enzyme was elevated. Subsequently he underwent testing that revealed advanced coronary artery disease and he had a bypass operation.
Even if we assume all the details of this story are accurate it is absolutely not a case of Apple Watch diagnosing an MI.
A high resting heart rate is not neccessarily an indicator of an MI and most MIs are not characterized by high heart rates. We have had the technology with wearables to monitor resting heart rate for some time and no one has ever suggested this can be used to detect MI.
The rate of false alarms is so high and the rate of failure to diagnose MI so low that this is a useless measure and should not provide any patient reassurance.
The writer of this story and the editors at 9-5 Mac should be ashamed of this misinformation.
Several other news sources have needlessly muddied the water on this question including Healthline and Fox News:
In clear cut cases the Apple Watch could make the difference between life and death,” says Roger Kay, president of Endpoint Technologies Associates. Because you wear the Apple Watch at all times, it can detect an early sign of a stroke or a heart attack, and that early indication is critical, he says.
And the Healthline article on the new Apple Watch also incorrectly implies it can diagnose MI:
The device, which was unveiled last week, has an electrocardiogram (ECG) app that can detect often overlooked heart abnormalities that could lead to a heart attack.
And if you are felled by a heart problem, the fall detector built into the Apple Watch Series 4 could alert medical professionals that you need help
Fox News and Healthline should modify their published articles to correct the misinformation they have previously provided.
And it is still true that although both Apple Watch and Kardia can diagnose atrial fibrillation the vast majority of the time acute heart attacks are not associated with atrial fibrillation.
Readers, please spread the word far and wide to friends and family-Apple Watch cannot detect heart attacks!
The results of the Apple Heart Study, were presented this morning at the American College of Cardiology Scientific Sessions amid intense media scrutiny. The AHS is a “prospective, single arm pragmatic study” which had the primary objective of measuring the proportion of participants with an irregular pulse detected by the Apple Watch who turn out to have atrial fibrillation on subsequent ambulatory ECG patch monitoring.
I and over 400,000 other Apple Watch owners participated in the AH study by downloading the Apple Heart Study app and self-verifying our eligibility.
My assessment is that we have learned little to nothing from the AHS that we didn’t already know. I’m also concerned that many patients suffered anxiety or unnecessary testing after being referred to urgent care centers, emergency departments, cardiologists or primary care providers and the results of these inappropriate referrals may never be determined.
Here is the study in a nutshell:
Participants enrolled by submitting information using the iPhone Heart Study app and none of their isubmitted nformation was verified.
An irregular pulse notification was issued to 0.5% of participants who were then contacted and asked to participate in a Telehealth visit with a doctor (who we will call Dr. Appleseed)
Only 945 of the 2161 who received a pulse notification participated in the first study visit.
Interestingly, Dr. Appleseed was empowered to send participants to the ER if they had symptoms (chest pain, shortness of breath, fainting/losing consciousness) It is not clear how many were sent to the ER and what their outcomes were but this flow diagram shows that 20 were excluded from further testing due to “emergent symptoms.”
Another 174 participants were excluded after finding out at the first visit that they had a history of afib or aflutter and 90 due to current anticoagulant use (both of these factors were exclusion criteria which gives us an idea of how accurate the information was at the time of participant entry.)
After all these exclusions only 658 ECG monitor patches were shipped to the participants of which only 450 were returned and analyzed.
This means of the original 2161 participants who were notified of pulse irregularity, the study only reports data on 450 or 21%. Such a low rate of participation makes any conclusions from the study suspect.
Of the 450 ECG patches analyzed only 34% were classified as having afib. Only 25% of this afib lasted longer than 24 hours.
After the patch data was analyzed, patients had a second Telehealth visit with Dr. Appleseed who reviewed the findings with the patient. Per the initial published description of the methods of the AHS (see here) Dr. Appleseed would tell the participant to head to the ER if certain abnormalities were found on the ECG.
Per the study description (apple heart study), Dr. Appleseed recommended a visit to the PCP for “AF or any other arrhythmia” detected by the patch:”
“If AF or any other arrhythmias have been detected in reviewing the ambulatory ECG monitor data, or if there are other non-urgent symptom identified by the study physician during the video visit that may need further clinical evaluation, the Study Telehealth Provider directs the participant to his or her primary health care provider”
At this point it seems likely that a lot of participants were instructed to go see their PCPs. Because as someone who looks at a lot of 2 week ambulatory ECG recordings I know that is the rare recording that does not show “other arrhythmias.”
Even more distressing is the call that participants would have received based on “the initial technical read:” I’m presuming this “technical read” was by a technician and not by a cardiologist. In my experience, many initial reads from long term monitors are inaccurate.
“If the initial technical read identifies abnormalities that require urgent attention (ventricular tachycardia or ventricular fibrillation, high-degree heart block, long pauses, or sustained and very rapid ventricular rates), then the participant is contacted immediately and directed to local emergency care or advised how to seek local emergency care.”
I wonder how many ERs had AHS participants show up saying they had been told they had a life-threatening arrhythmia? How much down stream testing with possible invasive, life-threatening procedures such as cardiac catheterization were performed in response to these notifications?
Overall, these findings add nothing to previous studies using wearable PPG technology and they certainly don’t leave me with any confidence that the Apple Watch is accurately automatically detecting atrial fibrillation.
Was more harm than good done by the Apple Heart Study?
We will never know. The strength of this study, the large number of easily recruited participants is also its Achilles heel. We don’t know that any information about the participants is correct and we don’t have any validated follow up of the outcomes. In particular, I’m concerned that we don’t know what happened to all of these individuals who were sent to various health care providers thinking there might be something seriously wrong.
Perhaps Apple and Stanford need to review the first dictum of medicine: Primum Non Nocere, First Do No Harm.
The Oura ring is a novel, multisensory device that claims to be able to distinguish sleep stages, including REM sleep,. I purchased one recently and after several months of evaluation and an extensive look at the data supporting it I have to say I am much more impressed with OURA’s hype, marketing and style than any useful or actionable information about sleep that comes from it.
The Oura website is full of pictures of cool people doing cool things wearing their Oura rings-like this guy
It’s also chock full of marketing blather which implies that somehow the ring will dramatically improve your sleep and your waking life.
We see every individual as unique: your state of health and wellness today, tomorrow, and days to follow. Getting enough restorative sleep has a profound impact on mental and physical health and performance. Your daily choices and rhythms define how well you sleep. With Oura, you learn your optimal times to move, eat and take a break to get that restorative sleep. Giving you actionable steps to improve your life opens a totally new universe of possibilities – be it for mental, cognitive or physical performance, or for beauty, health, and longevity.
A quick look at the OURA web site certainly conveys the sense that this is the slickest, most cutting edge personal wearable sleep and activity tracker one could purchase.
However, despite Oura’s tantalizing claims there is only one legitimate scientific comparison of the ring to the gold-standard of sleep evaluation, polysmnography (PSG). This was published in 2017 in Behavior Sleep Medicine and its full contents can be read here.
In addition, there is no published evidence whatsoever that changing one’s behaviour based on the various parameters that the ring produces will have any favorable effect on your sleep quality or health in general.
I’ll be quoting from that 2017 published paper which I think is a good, unbiased analysis and I’ll throw in some of my own observations throughout this piece.
How The Ring Works And What It Claims To Do
A good night’s sleep, everyone should know by now is incredibly important to optimal performance the next day. In addition poor sleep quality is linked to a whole host of pathologies (with causality yet to be proven for most.) Thus, I quickly purchased an OURAring after hearing Peter Attia rave about his ring.
OURA likes to promote the idea that it has some sort of special way of measuring sleep based on a combination of sensors.
The Oura ring and its proprietary algorithms are a combination of extensive scientific understanding, years of careful research and development work, and top-notch engineering. All insights and guidance you get are based on proven algorithms and verified knowledge. For example, Oura’s sleep staging algorithms were the first in the market that have been independently validated. The validation study was made by SRI International.
The OURA website notes that the ring is fitted with the following sensors to collect physiologic signals from your body.
NTC BODY TEMPERATURE SENSOR The Oura ring registers your body temperature reading every minute while you sleep. By comparing that value to values from earlier nights, it indicates your body temperature baseline and any variations from it.
INFRARED LEDS Measuring blood volume pulse directly from the palmar arteries of the finger.
3D ACCELEROMETER AND GYROSCOPE Detects the amplitude and intensity of your body movement, automatically recognizes that you’re active and tracks the time you were inactive during the day.
Ōuraring (Oulu, Finland) claims to use these physiological signals (a combination of motion, heart rate, heart rate variability, and pulse wave variability amplitude) in combination with sophisticated machine learning based methods to calculate deep (PSG N3), light (PSG N1+N2) and rapid-eye-movement (REM) sleep in addition to sleep/wake states.
After obtaining a sizing kit from OURA I selected my ring and within a few weeks it was delivered. I downloaded the free OURA iPhone app, charged the ring with the supplied USB charger, slipped it on my left ring finger and eagerly awaited my first night’s analysis.
Upon arising in the morning I opened the OURA app and visualized an entrancing display like the one below.
It’s a nice graphic summary of the night’s sleep with my minutes of REM, light, and deep sleep nicely quantified.
More graphs and more data are available by connecting to Oura’s online application which automatically syncs to your smartphone app.
Unfortunately, the app was telling me that I was awake for 109 minutes of the time I was in bed. Which was not correct. I was truly awake only for 10 minutes around 130 AM. This overestimation of my awake time has been a consistent error of the ring for my recordings. If the app can’t accurately track awake time all of its metrics are going to be inaccurate.
In fact, over several months of using the ring/app I have found little relationship between how I feel after sleeping versus how Oura has rated my sleep. There is even less correlation between the “readiness” score that Oura produces and how I feel during the day. Overall, I have found absolutely no actionable information from my months of using the ring.
One morning Oura gave me a “readiness” score of 68 and told me:
“Don’t push it. Your resting heart rate was above average, so you might not be fully recovered”
I felt great throughout the day. These recommendations in my experience are almost unversally inaccurate and useless.
Oura also makes recommendations on when it thinks you should go to bed. One time it told me I should go to bed at 7 PM. I have been ignoring its advice in this area.
Now I am just one individual and it is entirely possible there is something unique about my sleep that invalidates the ring’s accuracy. The ex-eternal fiancee’ tells me I’m a restless sleeper.
In fact, devices that rely on actigraphy tend to be fairly accurate at identifying when you are sleeping but not when you are awake which is the opposite of what OURA is doing in my case.
The SRI paper puts it this way
Compared to PSG, actigraphy has high sensitivity (ability to detect sleep) although specificity (ability to detect wakefulness) is lower(Marino et al., 2013; Sadeh, 2011), with a wide range of accuracy,depending on the amount of night-time wakefulness(Paquet, Kawinska, & Carrier, 2007),the algorithms used and the particular population studied(Van de Water, Holmes, & Hurley, 2011). Most importantly, actigraphy relies on a single sensor, an accelerometer, and thus it provides a measure of motion from which it predicts sleep and wake states. However, information about sleep stage composition, fundamental in studying sleep and sleep disorders,is not provided.
The Science Behind Oura’s Sleep Analytics: Detecting Sleep Stages
So what does the SRI paper OURA likes to quote as proving its accuracy say.
The paper is entitled “The Sleep of the Ring: Comparison of the ŌURASleep TrackerAgainst Polysomnography” and it was written by researchers at SRI international, a research consortium in Menlo Park, California with no ties to OURA.
Another paper which used to be touted on the Oura Ring website (but is no longer referenced on the site) utilized home PSG recordings and was done by an in-house OURA employee.
The SRI researchers studied 41 healthy adolescents and young adults with an average of 17 years and sleep data were recorded using the ŌURA ring and standard PSG on a single laboratory overnight. Metrics were compared using Bland-Altman plots and epoch-by-epoch (EBE) analysis.
EBE analysis showed that ŌURA accurately detected “light” and “deep” sleep in 65% and 51% of the epochs, respectively. It also accurately detected REM sleep epochs 61% of the time, with an overall overestimationof PSG REM sleep (by about 17 min). When the ŌURA ring misclassified PSG REM sleep, the algorithm classified the epoch as “light sleep” (76%) for the majority of the time.
These data suggest that the Oura Ring is virtually useless in telling you if you are in REM sleep versus deep or light sleep.
As the authors noted
Distinguishing sleep stages such as REM and N3 with non-EEG based systems has been challenging and is a goal of several commercial sleep-trackers, with mixed success.
Clearly, further work is needed to determine what combination of sensors might be used to optimally develop an algorithm that differentiates sleep stages sufficiently well to detect real differences or changes in healthy and clinical populations.
A look at the Bland-Altman plots really tells you how much variation there was in the PSG estimates of various parameters versus the OURA
The Bland-Altman plots show us how much the PSG time in REM differed from the Oura REM time for each individual subject. You can see that some individuals had considerable over-estimation of REM time whereas other had considerable overestimation of REM time.
Although OURA REM time was on average only 17 minutes higher than the PSG REM time this was because the marked overestimation of REM time in some (7 subjects over 30 minutes) was balanced by marked underestimation in others (9 subjects with over 40 minutes and one with 160 minutes).
Given that the average REM time was 92 minutes for most subjects there was a significant discrepancy between PSG. and OURA assessments.
OURA: Coin Flip For Detecting Awake
Oura ring was also pretty useless at identifying when you are awake
Overall, ŌURA had 96% sensitivity (ability to detect sleep), 48% specificity (ability to detect wake), 65% agreementin detecting “light sleep”, 51% agreementin detecting “deep sleep”, and 61% agreementin detecting REM sleep, relative to PSG
Like other sleep sensors utilizing actigraphy, Oura in most individuals can’t accurately differentiate between times when you are lying still but awake and when you are lying still and asleep.
The limitations of wrist actigraphy (see here) for differentiating sleep from wake are worse in those with insomnia:
With actigraphy, because sleep is inferred from lack of movement, subjects who are awake but lie motionless can be classified incorrectly as being asleep, and thus the technique is biased toward overestimating time to sleep, which may lead to incorrectly minimizing the severity of sleep disturbances. This may present a specific challenge for patients with insomnia, and may partially explain the limited validity of wrist actigraphy for estimating sleep onset latency..
There are multiple other issues and questions with the usefulness of the data that Oura provides that need clarifying before the ring can be considered useful.
For example the SRI paper found significant differences in results depending on which finger the ring was placed on.
Interestingly, we found that PSG-ŌURA discrepancies for “light sleep” and REM were greater on the ring finger compared to the other fingers, a result that was independent from the amount of PSG sleep fragmentation.Assuming that the main parameters that ŌURA uses to determine sleep stages are motion and optical sensor outputs, it is possible that the different blood supply among fingers maypartially explain these results. For example, it has been shown that SpO2 values differ between fingers as well as hands suggesting a finger-dependent difference in accuracy of the pulse oximetry signal (Basaranoglu et al., 2015).Further studies should confirm and better characterize the dependency of the PSG-ŌURA discrepancies on the ring position by having the same participants simultaneously wear different rings on different fingers.
The in-house Oura study also noted that results were more accurate on the non-dominant hand finger compared to the dominant hand but the Oura website makes no recommendation on which finger to use.
The other data that Oura compiles (heart rate, heart rate variability, temperature change, respirations) are clearly related to sleep cycles but Oura provides no evidence that these data or their proprietary algorithms to give you “readiness” or sleep quality scores are accurate or of any value.
Shold You Buy An Oura Ring?
If you are hoping to get improved analysis of your sleep quality I don’t think Oura adds anything to what is elsewhere available using cheaper wrist actigraphy devices.
The ring is expensive at 299$ and cannot accurately detect sleep stages.
Although most reviews you will encounter on the internet are wildly enthusiastic about Oura, they are likely biased and they provide no evidence that the unique aspects of the ring sensors provide useful information.
Would I buy it again?
I’ve misplaced my ring several times and I have to say that this distressed me immensely. Given that I think the sleep analysis is worthless this is hard to explain.
I think my attachment to the ring is due to a number of factors
It’s stylish and it mimics a wedding ring (which I otherwise would not have.)
I’m intrigued by some of the cardiovascular data it produces (night time heart rate and heart rate variability). Although currently I don’t think the data can guide me to healthier behavior, it’s possible that there is useful information in there somewhere. I hope to write a post on heart rate variability down the line. I’ve done research in this area and have some strong opinions on its value.
I’m curious to see if the respiratory rate data and the temperature data is of any value whatsoever.
So, the ring is best I would say for well-heeled,, self-hacking and self-experimenting techno geeks.
Apple claims that its Apple Watch can detect atrial fibrillation (AF) and appropriately notify the wearer when it suspects AF.
This claim comes with many caveats on their website:
Apparently it needs to record 5 instances of irregular heart beat characteristic of atrial fibrillation over at least 65 minutes before making the notification.
This feature utilizes the watch’s optical heart sensors, is available in Apple Watch Series 1 or later and has nothing to do with the Apple Watch 4 ECG recording capability which I described in detail in my prior post.
Failure To Detect AF
A patient of mine with known persistent AF informed me yesterday that she had gone into AF and remained in it for nearly 3 hours with heart rates over 100 beats per minute and had received no notification. She confirmed the atrial fibrillation with both AW4 recordings and AliveCor Kardia recordings while she was in it.
The watch faithfully recorded sustained heart rates up to 140 BPM but never alerted her of this even though the rate was consistently over her high heart rate trigger of 100 BPM.
The patient had set up the watch appropriately to receive notifications of an irregular rhythm.
Reviewing her tracings from both the AW4 and the Kardia this was easily diagnosed AF with a rapid ventricular response.
What does Apple tell us about the accuracy of the Apple Watch AF notification algorithm? All we know is the unpublished , non peer-reviewed data they themselves collected and presented to the FDA.
In a study of 226 participants aged 22 years or older who had received an AFib notification while wearing Apple Watch and subsequently wore an electrocardiogram (ECG) patch for approximately 1 week, 41.6% (94/226) had AFib detected by ECG patch. During concurrent wear of Apple Watch and an ECG patch, 57/226 participants received an AFib notification. Of those, 78.9% (45/57) showed concordant AFib on the ECG patch and 98.2 % (56/57) showed AFib and other clinically relevant arrhythmias. These results demonstrate that, while in the majority of cases the notification will accurately represent the presence of AFib, in some instances, a notification may indicate the presence of an arrhythmia other than AFib. No serious device adverse effects were observed
This tells us that about 80% of notifications are likely to be Afib whereas 20% will not be Afib. It is unclear what the “other clinically relevant arrhythmias” might be. If I had to guess I would suspect PVCS or PACS which are usually benign.
If 20% of the estimated 10 million Apple Watch wearers are getting false positive notifications of afib that means 2 million calls to doctor or visits to ERs that are not justified. This could be a huge waste of resources.
Thus the specificity of the AF notification is 80%. The other important parameter is the sensivitiy. Of the cases of AF that last >65 minutes how many are detected by the app?
Apple doesn’t seem to have any data on that but this obvious case of rapid AF lasting for 3 hours does not give me much confidence in their AF detection algorithms.
They do have a lot of CYA statements indicating you should not rely on this for detection of AF:
It is not intended to provide a notification on every episode of irregular rhythm suggestive of AFib and the absence of a notification is not intended to indicate no disease process is present; rather the feature is intended to opportunistically surface a notification of possible AFib when sufficient data are available for analysis. These data are only captured when the user is still. Along with the user’s risk factors, the feature can be used to supplement the decision for AFib screening. The feature is not intended to replace traditional methods of diagnosis or treatment.
My patient took her iPhone and Apple Watch into her local Apple store to find out why her AF was not detected. She was told by an Apple employee that the Watch does not detect AF but will only notify her if her heart rate is extremely low or high. I had asked her to record what they told her about the problem.
As I’ve written previously (see here) the Apple Watch comes with excessive hype and minimal proof of its accuracy. I’m sure we are going to hear lots of stories about AF being detected by the Watch but we need some published, peer-reviewed data and we need to be very circumspect before embracing it as a reliable AF monitor.
My first patient this morning, a delightful tech-savvy septagenarian with persistent atrial fibrillation told me she had been monitoring her rhythm for the last few days using her Apple Watch 4’s built in ECG device.
I had been eagerly awaiting Apple’s roll out since I purchased the AW4 in September (see here) and between patients this morning I down-loaded and installed the required iPhone and Watch upgrades and began making AW4 recordings.
Through the day I tried the AW4 and the Kardia on patients in my office.
Apple Watch 4 ECG Is Easy and Straightforward
The AW4 ECG recording process is very easy and straightforward. Upon opening the watch app you are prompted to open the health app on your iPhone to allow connection to the Watch ECG information. After this, to initiate a recording simply open the Watch ECG app and hold your finger on the crown.
Immediately a red ECG tracing begins along with a 30 second countdown.
Helpful advice to pass the time appears below the timer:
“Try Not to move your arms.”
“Apple Watch never checks for heart attacks.”
When finished you will see what I and my patient (who mostly stays in sinus rhythm with the aid of flecainide) saw-a declaration of normality:
Later in the day I had a few patients with permanent atrial fibrillation put on my watch.
This seventy-something farmer from Bowling Green, Missouri was easily able to make a very good ECG recording with minimal instruction
The AW4 nailed the diagnosis as atrial fibrillation.
We also recorded a Kardia device ECG on him and with a little more instruction the device also diagnosed atrial fibrillation
After you’ve made an AW4 recording you can view the PDF of the ECG in the Health app on your iPhone where all of your ECGs are stored. The PDF can be exported to email (to your doctor) or other apps.
Apple Watch Often “Inconclusive”
The AW4 could not diagnose another patient with permanent atrial fibrillation and judged the recording “inconclusive”
The Kardia device and algorithm despite a fairly noisy tracing was able to correctly diagnose atrial fibrillation in this same patient.
I put the AW4 on Sandy, our outstanding echo tech at Winghaven who is known to have a left bundle branch block but remains in normal rhythm and obtained this inconclusive report .
Kardia, on the other hand got the diagnosis correctly:
One Bizarre Tracing by the AW4
In another patient , an 87 year old lady with a totally normal recording by the Kardia device, the AW4 yielded a bizarre tracing which resembled ventricular tachycardia:
Despite adjustments to her finger position and watch position, I could not obtain a reasonble tracing with the AW4.
The Kardia tracing is fine, no artifact whatsoever.
What can we conclude after today’s adventures with the Apple Watch ECG?
This is an amazingly easy, convenient and straightforward method for recording a single channel ECG.
I love the idea that I can record an ECG whereverI am with minimal fuss. Since I wear my AW4 almost all the time I don’t have to think about bringing a device with me (although for a while I had the Kardia attached to iPhone case that ultimately became cumbersome.)
Based on my limited sample size today, however, the AW4 has a high rate of being uncertain about diagnoses. Only 2/3 cases of permanent atrial fibrillation were identified (compared to 3/3 for the Kardia) and only 4/6 cases of sinus rhythm were identified.
If those numbers hold up with larger numbers, the AW4 is inferior to the Kardia ECG device.
I’d rather see the AW4 declare inconclusive than to declare atrial fibrillation when it’s not present but this lack of certainty detracts from its value.
What caused the bizarre artifact and inconclusive AW4 tracing in my patient is unclear. If anybody has an answer, let me know.
We definitely need more data and more studies on the overall sensitivity and specificity of the AW4 and hopefully these will be rapidly forthcoming.
For most of my patients the advantages of the AW4 (assuming they don’t already have one) will be outweighed by its much greater cost and we will continue to primarily utilize the Kardia device which will also allow me to view all of their recordings instantaneously in the cloud.
Note. The original version of this post had the wrong ECG tracing for the first “inconclusive” AW4 recording of a patient with permanent atrial fibrillation. H/t to discerning reader Vignesh for pointing this out months after the initial posting.
I described in detail in March (see here) my early experience in utilizing AliveCor’s KardiaMobile ECG device in conjunction with their Kardia Pro cloud service to monitor my patient’s with atrial fibrillation (afib). Since that post the majority of my new afib patients have acquired the Kardia device and use it regularly to help us monitor their afib.
This capability has revolutionized my management of atrial fibrillation. In those patients who choose to use AliveCor there is really no need for long-term monitors (Holter monitors, Zio patches, cardiac event monitors) and no need for patients to come to the office to get an ECG when they feel they have gone into afib.
When one of my Kardia Pro patients calls with symptoms or concern of afib, I quickly pull up their chart at Kardiapro.com and review their recordings to determine if they are in or out of rhythm. Most treatment decisions can then be handled over the phone without the need for ordering a monitor or an emergency room or office visit.
One 24 hour period will suffice to show how important KardiaPro is now to my management of my patients with afib
A Day In The Afib Life
Tuesdays I spend the day working in the heart station at my hospital. Typically, on these days I will supervise stress tests, read ECGs and echocardiograms, perform TEES and electrical cardioversions. On a recent Tuesday I had 3 patients scheduled for cardioversion of their atrial fibrillation.
The day before one of these patients called indicating that he suspected he had reverted back to normal rhythm (NSR) based on his Kardia readings. He had had a prior cardioversion after which we know (thanks to daily Kardia recordings) he reverted to afib in 5 days. Subsequently we had started him on flecainide, a drug for maintenance of NSR and scheduled him for the cardioversion.
Not uncommonly after starting flecainide patients will convert to NSR but if they don’t we proceed to an electrical cardioversion.
I logged into KardiaPro and reviewed his dashboard and sure enough his last two ECGs showed sinus rhythm. I congratulated him on this and we canceled his cardioversion for the next day, saving the lab the time and expense of a cancellation the day of the procedure. The patient avoided much stress, time and inconvenience.
It is important to note that in this patient there was no great jump in heart rate with afib compared to NSR. For many patients the rate is much higher with the development of afib and this is often detected by non ECG wearable monitors (like an Apple Watch.) But for patients like this one, an ECG is the only way to know what the rhythm is.
A second patient with afib who had elected not to acquire an AliveCor ECG device showed up for his cardioversion on Tuesday and after evaluating his rhythm it was clear he had spontaneously reverted back to NSR. Prior to my adoption of KardiaPro this was a common and scenario.
The third scheduled cardioversion of the day showed up in afib and we successfully cardioverted him back to NSR. I had not addressed utilizing AliveCor with him. Prior to the procedure he asked me about likely outcomes.
My standard response to this question is that we have a 99.9% success rate in converting patients back to NSR at the time of the cardioversion. However, I can’t predict how long you will stay in NSR after the cardioversion. NSR could last for 5 days or it could last for 5 years. Adding medications like flecainide or amiodarone can significantly reduce the risk of afib recurrence after cardioversion.
At this point he asked me “How do I know if I am in afib?” Whereas many afib patients immediately feel bad and are aware that they have gone out of rhythm, this man like many others was not aware.
Prior to AliveCor my answer would have been to check the pulse daily or look for evidence of high or irregular heart rates on BP monitors or fitness wearables. This scenario provided a wonderful opportunity to test the AliveCor’s accuracy at detecting AF in him. I pulled out my trusty AliveCor mobile ECG and prior to the cardioversion we made the recording below
After the cardioversion we repeated the Alivecor recording and the rhythm (AliveCor’s interpretation) had changed from afib to NSR.
Needless to say, this patient purchased a Kardia device the next day and since the cardioversion he’s made a daily recording which has confirmed NSR. I just logged into Kardia Pro and sure enough he made a recording last night and it showed NSR.
Later in the week I received a call from a patient I had electrically cardioverted a few days earlier. His Kardia device had detected that he had gone back into afib.
I logged into my Mac and saw his KardiaPro chart below.
With perfect precision KardiaPro had verified NSR after the cardioversion lasting for 36 hours. For some reason after dinner the day after the cardioversion, the patient had reverted back to afib. This knowledge greatly facilitates subsequent treatment and eliminates the need for in office ECGs and long term monitors.
Utilization of the Kardia device with the Kardia Pro monitoring service has proved for me to b a remarkable improvement in the management of patients with afib. Managing non Kardia afib patients feels like navigating a forest with a blindfold.
The improvement is so impressive that I find myself exclaiming to my assistant, Jenny, several times a week “How do other cardiologists intelligently care for afibbers without AliveCor?”
I have a few patients who balk at the 15$ per month charge for Kardia Pro and ask why the device and this monthly charge aren’t covered by insurance or Medicare. Given the dramatic reduction that I have noticed in my use of long-term monitors as well as office and ER visits in this population, CMS and third-party insurers would be wise to explore Kardia monitoring as a more cost-effective way of monitoring afib patients.
N.B. I realize this post appears to be an unmitigated enthusiastic endorsement of a commercial product which is quite uncharacteristic for the skeptical cardiologist.
One might wonder if the skepcard is somehow biased or compensated for his endorsement of Kardia.
In all honesty, this sprung from my love of the device’s improvement in my afib management and I have received no payment, monetary or otherwise from AliveCor and I own none of their stock (and I’m not even sure if it is on the stock market.)