Cardiologist Questions Accuracy of ECG Function in Apple Watch 4

Of all the features of the Apple Watch 4, probably the most-discussed is its ability to perform an electrocardiogram, or ECG. An ECG (also called an EKG) is typically performed by putting electrodes on the skin to detect the electrical changes that arise from the beating of one’s heart. The Apple Watch 4 will include both the ability to perform an ECG and to detect irregular heartbeats, but one cardiologist has questioned whether the application is sufficiently accurate to perform these tests safely.

First, one point to keep in mind: Neither the ECG nor the irregular heartbeat pickup software is currently baked into the product. Apple has not given a date for when either will arrive. The second point — the one hasn’t been as discussed — is that some doctors are themselves cautious about what kind of information these apps will provide and how useful it will be. They have good reason to be.

Any time you test for a medical condition (and the Apple Watch 4 is passively monitoring for one), you have to address the risk of both false positives and false negatives. A false positive is a test result that says you have a problem when you actually don’t. A false negative is a test result that states you don’t have a problem when you actually do. Both can result in death or, at the very least, sub-optimal or incomplete medical treatment for an ongoing condition.

Most of the coverage on the Apple Watch 4’s new medical monitoring dug into the question of what Apple offered, but not the evidence presented to the FDA to gain clearance to offer the capability in the first place, as a new story at HealthNewsReview points out. Apple gathered data from 588 people for its ECG app and 226 people with an atrial fibrillation.

For the ECG app, the FDA received data showing that the Apple Watch couldn’t classify about 1 in 10 rhythms. Of the rest, 98.3 percent of the time, people with atrial fibrillation were correctly detected as having it, while 99.6 percent of people were correctly identified as not having an AFib if they didn’t have one.

The irregular heartbeat monitor wasn’t as accurate. In that mode, 226 patients were followed using an ambulatory cardiac monitor for six days. 41.6 percent of patients had an episode during that time. Of the 41.6 percent of patients who had an episode, the Apple Watch 4 caught an event in 78.9 percent of cases. But there are two important caveats to both of these metrics: The data hasn’t been peer-reviewed, and in these cases, the clinicians already knew which people had AFib issues and which did not. A percent or two of incorrect assessment scales to encompass millions of people when you consider the size of Apple’s worldwide market.

“The big problem with this conclusion is that this population has a prevalence of AFib that is probably 100-fold larger than Apple’s target market,” University of Michigan cardiologist Venkatesh Murthy, MD, told Health News Report. “This is not good. However, the major caveat here is that we are still lacking most of the information needed to be sure how this experiment was done, so we really are just guessing.”

There are also questions about the overall process used to verify health software in the first place. Apple is part of a program to speed certification of health software, but the FDA’s rules for approving devices in this context appear to be quite weak. The FDA goes so far as to state that “demonstration of [safety and effectiveness] does not necessarily require a clinical investigation/study.” That’s a surprising declaration from the organization in the wake of high-profile flameouts like Theranos, in which a company valued at hundreds of billions of dollars essentially lied about its products and their ability to transform the market for blood tests for several years until its entire business was proven to be founded on lies and the company imploded.

It’s an open question whether this kind of monitoring is useful to most people or not. This isn’t a trivial question, either, when the price of health care is skyrocketing. It’s not so much whether the Apple Watch 4 could detect a potentially life-threatening problem, but whether it’ll send people to expensive specialists (or even increasingly expensive PCPs) unnecessarily. The problem with pushing back against the monitoring narrative is that there are already real-life instances of features like this potentially saving lives. Discussions about the perceived need for care, whether people should be heading to the doctor in any given situation, and the long-term inflation of care pricing in the American medical system (all enormously complex topics to begin with) are even more difficult to have when, on the one hand, you have a bunch of dry facts and figures and on the other you have persuasive personal essays with headlines like: “How Apple Watch saved my life.” So it has ever been, in every era. But that doesn’t make the points unimportant.

Buy an Apple Watch 4 if you like and use the ECG and AFib monitoring capabilities if it gives you some peace of mind, but never make the mistake of thinking it’s a cardiologist — and keep an eye on how accurate the software turns out to be once it’s in-market and better tests can be done. That’s not as good as testing the device in blind studies to begin with, but it’ll answer some questions about just how accurate the feature is.