"You should buy Apple stock. This probably saved you."
A doctor reportedly told the patient who went to the clinic after his Apple Watch detected atrial fibrillation (Afib for short). This story happened the day after Apple Watch released its ECG app. You must have heard this news story and others like it. One might think, this is great. Now my Apple watch can tell me when I have Afib. I will go to the doctor and get treated and I will be fine.
Let’s take a step back and look at this problem. Atrial fibrillation is a common type of irregular heartbeat, and its occurrence increases with age, affecting about 3% of men and 2% of women aged 65 to 69 years and about 10% of adults 85 years and older. It is much less frequent in folks who are younger than 65 years of age. Folks with Afib are at more than 5 times the risk for stroke, compared to those who do not have it. Approximately 20% of patients who have a stroke have associated Afib. That is when they first realize that they have Afib.
Electrocardiogram (ECG for short) was discovered about a hundred years ago. It is a graph derived from the electrical activity of the heart, detected by placing electrodes on the skin. The typical ECG that you get in the doctor’s office is a rough 3D representation of the heart’s electrical activity, which is based on the health of the “wiring” inside the heart muscle. The more recent inventions can measure this activity using wireless sensors on your skin, like the Apple Watch. It has been tested to be fairly accurate, like the ECG done in the doctors’ offices which is still considered the gold standard.
People diagnosed with Afib, need to be treated for rate control if the heart rate is fast, rhythm control with medicines and/or electric shock to bring it back to normal rhythm, with a blood thinner medicine to reduce their risk of having a stroke in the future and modifying the risk factors that may be causing Afib. OK, so far so good. We could screen everyone who wears such a device and if they have Afib, we treat them with medicines or electric shock and then put them on blood thinners. That is how you can save people’s lives by decreasing their risk of having a stroke.
This is awesome! Fitting a briefcase size ECG machine into a watch is an engineering marvel (even if it records just 1 lead). The user interface that Apple is famous for is great. The convenience and the ease of use is awesome. The connectivity and being ubiquitous is terrific. The accessibility of a medical grade device on your wrist which is user friendly, portable, light and cute is amazing!
Then what is the problem with using Apple Watch or other similar devices to detect Afib?
Challenge #1: The prevalence of Afib is less than 2% in people who are younger than 65. When you have a disease with low prevalence like Afib, and you use a test like ECG (or Apple Watch), you may get more false positives than true positives. In other words the performance of the ECG test depends on the prevalence of Afib in a given population. The US Preventive Services Task Force (USPSTF for short) is a body of experts that looks at research on diseases and tests to evaluate them. In its Conclusions and Recommendations, the USPSTF in 2018 stated that “the current evidence is insufficient to assess the balance of benefits and harms of screening for atrial fibrillation with ECG.” What?! They are saying Apple Watch and other such devices, including the ECG done in the doctor’s office is useless for Afib screening in the general population!?
Challenge #2: Not everyone that is diagnosed with Afib will benefit from blood thinner medicine! When someone has Afib, their doctor evaluates other risk factors for stroke and decides if the risk is high enough to benefit from treatment with blood thinners. The reason for that is – blood thinners have side effects. They can cause serious bleeding in the gut, brain and other places in the body, which may sometimes lead to death in a small number of patients. Yikes! So, the treating physician has to weigh if the benefit from the treatment is bigger than the risk from treatment. If the benefit is significantly larger than the risk, then those patients will be started on blood thinner medication.
Challenge #3: So far studies have been done to asses patients using the gold standard ECG to confirm the diagnosis. A typical new Afib patient will have a symptom like dizziness, palpitations, and shortness of breath or chest pain from the Afib. They then go to the doctor’s office or emergency department, where they get the ECG done. That is how they usually get a diagnosis of Afib. What if the patient is not having any symptoms and the monitoring detects periods of Afib? How long should the period be to become a risk factor for stroke? Do many short intermittent periods carry the same risk as continuous Afib? What if it lasts for a few seconds or minutes? We don’t have solid data on these issues. The ability to detect asymptomatic periods of Afib has complicated the diagnosis and treatment process.
Challenge #4: The following risk factors have been implicated in causing Afib: advancing age, high blood pressure, obesity, European ancestry, diabetes, heart failure, ischemic heart disease, hyperthyroidism, heavy alcohol use and enlarged heart and chronic kidney disease. Many of these risk factors can also cause strokes and heart attacks, with or without Afib. In fact, Afib was found to be associated with heart attacks too. Now blood thinners are prescribed to prevent stroke and heart attacks. So then, is Afib a disease or the complication of a disease or a risk factor for disease? It is actually all three. The common pathway for all these conditions could be diseased and narrowing blood vessels affecting the function of the organ involved. So the relationship between risk factors and outcomes is not linear, but a complex web. It is more like a tree with branches and roots with branches and roots coming from the branches. – All in an interconnected web!
Challenge #5: When someone detects Afib, they go to their doctor’s office or the emergency department - mostly to the emergency department, because now they are scared and anxious. OK, they get checked out, treated if needed and then are recommended follow-up testing and care. Let’s say the Afib is real and you do have enough risk factors to get started on stroke prevention medication. If you have insurance and your doctor is getting paid per visit, that’s great. But what about the hospital that may have lost a potential future stroke patient! This is a big challenge for the fee-for-service business model of healthcare delivery, which is where majority of the provider market currently is. Prevention sounds great in principle but when the rubber hits the road, it does not “work” in a fee-for-service model. On the other hand, value-based care providers take prevention seriously as it affects their bottom line and where the benefit to the patient is also a benefit to the healthcare provider company.
In conclusion, healthcare is very complicated. It is awesome that technology companies like Apple and others are trying to solve problems in healthcare. I applaud the effort in this endeavor. They bring ingenuity, great minds, outsider perspective, cool technology and money to invest. It is great that Apple Watch has shone a light on Afib and brought increased awareness about this condition. Solving technical problems in healthcare is the easy part. Making a solution actually change patient outcomes is the hard part. There are so many other ingredients that are needed. Cassie Kozyrkov of Google, uses the analogy of pizza making to decision science. In this new ECG technology, we found a new tomato. Now, how do we get to the pizza, the final outcome? There are many more ingredients and processes and pieces of equipment needed to get there. May be it is time for the large technology companies to take the bull by the horns. Buy the pizza shop!