I’ve been evaluating the ability of a mobile hand-held ECG device called AfibAlert to detect atrial fibrillation for the last few weeks.
I found that the device made very reliable and consistent recordings of cardiac rhythm and did a reasonably good job of detecting atrial fibrillation (afib).
The device came in a plastic case with a USB cable for uploading recordings and two metal bracelets which attach to electrodes and provide an alternative recording method.
The device itself is about 6 by 3 by 1 inch.
Recordings are made by placing your thumbs on the silver/siver chloride electrodes
After a few seconds the display in the center will give heart rate and after 45 seconds the
device will make a decision about your rhythm:
If it diagnoses normal sinus rhythm a green check appears and if it diagnoses afib a red telephone appears.
If it is confused you get yellow circular arrows.
As the maker of the device explains:
Lohman Technologies’ patented algorithm analyzes the patient’s heartbeat and the appropriate icon illuminates to show what action is needed. AfibAlert’s® algorithm was validated against 51,000+ ECG strips from the MIT-BIH Atrial Fibrillation Database with known diagnosis. The Afib monitor’s results were excellent, with 94.6% accuracy in detecting the presence of arrhythmias. Each recording produces a 45-second diagnostic quality ECG rhythm strip
The device I tested does not allow you to immediately see the ECG tracing. The recordings are uploaded to a PC via USB cable and then can be viewed as a PDF document.
I made 17 recordings on patients in my office one day. The age range was 50 to 93 years and most patients were able to rapidly and easily grasp the device with thumbs appropriately positioned to make interpretable recordings.
Only 2/17 came back. yellow. In both cases, I repeated the recording and the device was able to make the correct diagnosis. Twice I got the yellow signal on an elderly, partially blind patient who had trouble keeping his thumb on the electrode.
In 15 cases of normal sinus rhythm the device correctly identified NSR.
In one case of atrial fibrillation the device correctly identified atrial fibrillation.
In one case of SR with
APCs the device
incorrectly identified afib
Overall the device correctly classified 88% of the tracings. This was superior to the device I normally utilize ( AliveCor/Kardia mobile ECG) in head to head comparison (I’ll present this comparison in a subsequent post).
My bullet points on the AfibAlert device:
-5 stars for Ease of Recording
-5 stars for Quality of recording
In all cases that uploaded, the recordings were very clear and free of artifact. The device did not upload yellow signal events and I presume more artifact is present in these recordings.
-2 stars for Convenience.
I found the software and uploading to be very awkward and slow. The company indicates new software soon to be released along with the ability to interface directly with iPads or smartphones that hopefully will improve this factor.
The inability to instantaneously view the ECG tracings means I cannot use it in my office to screen patients for arrhythmias. However, if a patient is solely using it to determine if afib is present or absent, this information is available right away.
-3 stars for Accuracy.
It does a reasonable job of identifying the patient who is in normal rhythm versus one in atrial fibrillation.
However, like AliveCor and other devices which strictly look at the variability of the pulse, it can be easily fooled by premature beats, especially when they are frequent, and inappropriately classify these as afib resulting in false positives.
In addition, when afib rates are very slow and thus much less variable it is likely AfibAlert will incorrectly classify them as normal thus resulting in false negatives.
False Negatives and False Positives
False negatives result in delayed diagnosis of afib. Patients will be falsely reassured that their rhythm is normal when it is not.
False positives result in needless anxiety and testing/treatment.
If afib monitoring devices are to be successful they have to have a very low frequency of both types of inaccuracy.
The solution to inaccuracy of interpretation, of course, is to have a cardiologist over-read the tracings.
AfibAlert recordings are available online for review by your personal physician after being uploaded. This requires your physician to have an account with AfibAlert. There is no capability for having the recordings over read by an online cardiologist for a charge.
As far as I can tell the device is only available for purchase in the US and only on the AfibAlert website.
Interestingly, you cannot purchase AfibAlert without a prescription from a physician.
Why this is mandated for AfibAlert and not AliveCor is a mystery to me.