I'm a Heart Doctor and Wary of the New Apple Watch
medium.com
medium.com
I agree with this statement generally, but in this specific case it seems tenuous. If the iWatch’s 1-lead EKG detected a-fib, the next course of action would be a more reliable diagnostic, specifically a proper 12-lead EKG interpreted by a qualified medical professional, perhaps followed by a Holter monitor to record 5 or more leads over a 48+ hour period.
A 12-lead involves attaching ten electrode stickers to a patient which are then wired up to a backpack-sized monitor which passively reads the difference in electrical activity between specific electrodes. It’s a simple non-invasive procedure that’s performed a hundred thousand times a day in the US. In fact, it’s so simple and low-risk that I as a lowly EMT am allowed to perform it. (To drive this point home, the state of California does not allow me to prick a patient’s finger to get the drop of blood required to ascertain blood sugar. I need a paramedic to do that for me.)
It’s absurd for this doctor to suggest an asymptomatic young person could be prescribed anticoagulants for a-fib “diagnosed” by an iWatch. I can understand doctors expressing opposition to “self-diagnostic medicine” but to make these arguments about a 1-lead makes as much sense as arguing that people shouldn’t be able to buy and operate a blood pressure machine without a prescription.
A 12-lead ecg can diagnose whether you have a-fib at the time of the ecg. The question for the doctor is whether you may have “paroxysmal” ie intermittent a-fib. So what if the Apple Watch said ‘a-fib’, but the 12-lead is negative. Most doctors are probably going to end up ordering a holter- monitor, to monitor the patient’s rhythm over 24-72 hours. Some proportion of those patients may have very short runs of a-fib, and would likely have remained blissfully ignorant (and well) without the Apple Watch. Instead, they may well end up anti-coagulated, because if that patient is not anticoagulated and gets a stroke later ( which may or may not be due to a-fib), the doctor is going to get sued. Whereas if the doc starts the patient on anticoagulants, and tells the patient that, “ there’s a risk of bleeding, but...”, that is much less likely to end up in a lawsuit.
If you are going to act on the information, then you must weight the benefit of early detection against the damage caused by unnecessary exams and treatment when there are false positives.
False positives are specially problematic when screening a large number of healthy people.
Whether you class these as false positives (because the existing test suite doesn't detect them) or false negatives (because the existing test suite should have detected them) is perhaps a matter for debate. The practical outcome is more people living healthier lives because their persistent malaise has finally been identified.
At first I was wondering how scheduling someone for a long-term ECG would cause damage, then I remembered that this feature will only become available in the USA soon. Given that the USA is a country where unexpected health care can destroy a person's life through either bankruptcy or the curse of being a woman/person of colour, yes I agree that detecting a previously undetectable condition could cause damage.
Time and data will surely tell if this type of ECG screening is helpful or harmful on a population level. Certainly, some people will get a secondary ECG indicating intervention is required, and the intervention will improve their lives. Others may have a negative result from the intervention, and may never have found their way into this intervention without this screening. For some, this screening indicating a problem but a full screening indicating nothing will drive persistent anxiety.
It is absolutely not the case that more medical intervention is an unqualified good, even ignoring the economic costs.
More data is useless if your data is mostly noise and if as a result error rate in treatment and costs rise. Add to this the strong incentive for doctors, especially in a profit-driven healthcare system to suggest action, this could easily lead to the rise of unnecessary treatment of benign abnormalities in otherwise healthy people.
This is also not speculation. Ever since the rise of diagnostics in medicine overtreatment has become more and more common. This study[1] estimates that almost a fifth of all procedures in the US are not necessary.
You are making a causal connection of rise in diagnostic tools => overtreatment. What is your evidence for that, apart from overtreatment increasing? I think sea levels have risen since then too, are they also an effect of too much diagnostics?
Clearly, if the data was just noise, people would eventually stop caring. But you'd have to be pretty cynical to think one of the major tech companies of the world releases a diagnostic device that has an S:N ratio that high.
A more likely explanation for the overtreatment is simply that people are getting richer, and society is providing more care -- to a fault. I know this first hand, I have had invasive surgery done that I would've been better without -- because the doctor thought it maybe might be a good thing for the future. Spoiler: it worsened the problem.
I think if it were the case that we all suffer from a bit of a-fib every now and then, and that these diagnostic tools will give false positives for this reason, then that would be news to the medical community and thus progressing our understanding. People would be told, "your watch is telling you this but we know better now." Apple might update their detection algorithm. It's such a small price to pay for the obvious advantage of preventing loss of life.
On the other hand, maybe more data could be harmful? You might be interested to read a classic article (Bergman, 1967, [“The morbidity of cardiac nondisease in children”](http://dabamirror.sci-hub.tw/4363/be7b5e26e14c6389108fbdc70f...)
It discusses the harm to children of their parents believing their child has a “heart problem” when they have a completely benign murmur - which raises the issue of whether you should tell a parent there is a benign heart murmur when that information might actually be harmful - and whether that risk of harm is sufficient to override the parent’s right to know.
Another example will soon be genetic testing for risk of Alzheimer’s disease - if, at present we have no way of reducing the risk, is it fair to give people that information, when we know based on many studies that such knowledge lowers patient’s quality of life?
I think free adults should have the right to do all sorts of things that others think might be harmful - but it also seems reasonable (as this article does) to make people aware of the risks of “more data”, as well as the assumed benefits
If you want to argue against gathering information in this way, then perhaps phrenology serves as an interesting example. We know now that most likely, it was a croc. The use of phrenology was to draw a boundary around a certain ethnicity so that it could be upheld above all other ethnicities, and so it is hard to say that the data collection itself was the problem. Sweden is having a similar but opposite problem right now -- assault and sexual assault perps are predominantly Arabic immigrants, for one reason or another -- maybe it's in part Middle Eastern culture, almost assuredly it's a lot about socioeconomic factors, but nonetheless, it is true.
The powers that be have decided that the news should not mention ethnicity or citizenship for fear of racist backlash, they have decided that the annual crime statistics reports should not mention ethnicity or citizenship, again citing fear of racist backlash. I now invite you to consider, what will the population think then, if there are no facts to refer to, nothing to dissuade the racist's viewpoint that "Muslims are bad people?" Facts are sometimes inconvenient or inflammatory, but that doesn't make them less true.
For your case, I think the problem is not one of information gathering. Doctors should still be examining the kids, I think you agree -- the problem is giving the wrong impression to the parents, or not knowing that it is in fact the wrong impression.
Heart risk is serious and by itslef warrants a GP visit or at least a question. Yet a recreational health monitoring looks similar to someone's reading a GP's handbook for fun - suddenly lots of descriptions seem to apply due to uninformed perceptions.
Either way, if one has reasons to believe being at heart risk, watch is not the first thing to buy to be pro-active/preventive.To have the watch does not mean to take care of the risk. Just like having a fitbit does not mean one is actively fit.
Maybe a similar example is with a dog bite. If you say 'no it was my dog and they are not rabid' then doctor says 'ok no rabies shot needed'. If you say 'it's a wild dog (or friends dog)' and you don't know if it's rabid doctor thinks 'rabies shot needed'. Even if the actual chance of rabies (in your friends dog) is quite small (if the friend doesn't know if the dog has shots I mean).
More info is not only not always helpful but comes with risks as well. If not we would monitor things more frequently than we do and the simple scale would lower the cost.
Are they already condemning people to unnecessary treatment on the off-chance that treatment is helpful?
Is there scope here for the AMA to refine procedures to suggest holding off on medication when a condition is only discovered by a persistent monitoring system such as Apple Watch?
Even if the over-medication issue increases, are the live saved via early detection of equal value to the lives damaged by unnecessary medication? If there are more lives saved than damaged, is there a net benefit to society, or are we only interested in the lives lost through action and not at all concerned about the lives lost through inaction?
I stand by my argument that a GP prescribing an anticoagulant based solely on an iWatch having detected a-fib is grossly negligent (just as—in keeping with my earlier comparison—he would be if he prescribed an anti-hypertensive to a patient who stated they had a BP of 180/120 at home one time).
I’m not entirely sure why you docsplained that 12-leads cannot bend spacetime and that a Holter monitor would be likely be ordered to detect intermittent arrhythmias over a longer period of time, but thanks. Can I claim CE hours for this?
That may not materially change the public health calculation, or it may. But that's the calculus -- not that a GP would blindly prescribe, just that increasing the top of the funnel is not necessarily a good thing. (Plus, extra cost.)
Positives and negative impact of additional testing must be measured for optimal decisions.
* numbers to make the math easy
I’m cautiously optimistic but recognize that we need more data because evidence-based medicine is all about testing hypotheses. The hypothesis here is that continuous 1-lead monitoring will provide a net benefit to the population employing it. Maybe it will and maybe it won’t; Either way I think it’s pretty cool that we’re about to amass a ton of relevant data from which we’ll be able to derive an answer.
Screw this "doctor" and their FUD.
I think you’re missing the argument made (fairly poorly, IMO) in this article: that preventive medicine often isn’t worth the money for society.
For relatively rare diseases, false positives typically outnumber true positives by a wide margin. If that’s the case for this Apple Watch, that means many users will needlessly be stressed thinking “I’ve a heart problem”, something that will cause some loss of QALY’s (https://en.wikipedia.org/wiki/Quality-adjusted_life_year) and will trigger tests that will take resources that may better be spent elsewhere.
That’s for society, though. Richer I ndividuals may not care about the extra costs, and people doing lots of checkups may be less stressed about false positives.
What worries me is that a 1-lead ECG is not worth much. If a 12-lead is only 90% accurate, a 1-lead is bound to be much worse.
The fact that medicine hasn't studied whether or not healthy hearts are regularly doing this because sensors were inconvenient previously, doesn't mean we need to get rid of the sensors. It means that both the presentation of the software should have some sort of feedback loop from clinical research that should emerge as a consequence of that data now... existing.
That will, on the other hand, increase false negatives. I think that will defeat the whole purpose of putting the sensor in the watch.
> medicine hasn't studied whether or not healthy hearts are regularly doing this because sensors were inconvenient previously
Medicine has studied this. It is just that this specific kind of diagnosis is subjective and there are a lot of debate.
The main point, from what I understood, the author is arguing is that the one-lead EEG of the Apple watch is prone to higher false positives, and users should take it with a grain of salt.
In fact, while this may at first lead to a bunch of hypochondriac-like behaviors, people's fascination with themselves will quickly teach them to not be alarmed.
This is completely grounded in science. And we should be weary of going to the doctor too often:
>Medical Errors Are Third Leading Cause of Death in the U.S.
1. https://www.usnews.com/news/articles/2016-05-03/medical-erro...
They eventually told me to get a Kardia so I could check it whenever I wanted to.
After a few weeks of checking every time I thought something was going on and periodically sending the worst examples to their analysts, I realized it was only ever pvcs, and it dramatically lowered my stress levels and made the episodes much less severe and shorter lasting. It also helped me figure out my triggers (caffeine) and helped me realize that going on a brisk walk for 15 minutes pretty much always stops them.
I wouldn’t say the Kardia saved my life or made me healthier, but it absolutely helped me get a handle emotionally on what was going on with my heart and improved my quality of life. I don’t carry it around any more, but I definitely think it’s worth getting (or the new watch) if you have palpitations and want to get a better idea of what’s going on.
Is it worth it to give 1000 people anxiety, tests, medication in order to provide benefit to 10 people or even 100? (those numbers are made up to illustrate and are fictitious).
Take a FitBit Charge HR 2. The one I have is consistently within 2-5% of hospital & doctor office pulse readings. That doesn't mean it's perfect in all circumstances, but that it's observations haven't been discredited with evidence and there's little/no counterexamples to show that a device I wear is somehow "inaccurate."
Another doctor claimed home blood pressure cuffs, products strictly-regulated by the FDA, are "widely inaccurate." This also seems binary-thinking FUD and wrong, because, anecdotally, I checked repeatedly and have an accurate consumer blood-pressure meter.
This is what happens when a profession encourages people to invest their egos in a preordained outcome.
As well, if you have peers who have Apple Watch/fitbit, there are tons of apps you can use to "compete" with them.
When we were bootstrapping with a similar idea as Apple Watch at Aidlab, all cardiologist were looking at us like we were crazy. Well, seems we were talking with wrong doctors, and we should go straight to Dr. Ivor Benjamin.
Relevant snippet: ‘The potential gains of screening are opposed by reduced QoL related to diagnostic biopsies, burdensome cancer treatment, including radical prostatectomy (RP), radiotherapy (RT), and androgen deprivation therapy (ADT), and the frequent long-term treatment-related adverse events, including erectile dysfunction (ED), urinary incontinence (UI) and bowel dysfunction (BD) [12–14]. In addition, RP has a small risk of peri-operative death [12, 13].
Finally, since a substantial fraction of PCas shows late onset and slow progression, overdiagnosis and overtreatment are common consequences of screening, especially when it is performed repeatedly or in men with relatively short remaining life expectancy due to age or life-shortening co-morbidity. Overdiagnosis and overtreatment are difficult to assess in empirical studies, because it would require a lifelong follow-up of men randomized to no screening and various screening options in a study without migration bias.“
I get that screening may not be worth the cost, but if the screening is effectively free... then it seems the problem is more in how we respond to the answers.
What it boils down to is the rate of false positives versus the rate of undetected positives (as opposed to false negatives). If, for example, the Apple Watch in use by 1000 people detects 3 people who with a-fib who previously were not aware of their condition, we might say, "excellent!" But on the other hand if it turns out that one of those people didn't have a-fib, was put on medication and suffered complications, we now have 1 person needlessly suffering who wouldn't have been suffering before and 2 people who are in no better or worse shape than they would otherwise have been. If the regular detection rate of a-fib leads to about 1 person in 1000 being granted an improved quality of life, has the a-fib detection program been useful or harmful?
Then it gets worse: what if the test we use to "diagnose" a condition have a high false positive rate? One example is screening for lung cancer: the screen itself has a 35% false positive rate, but the condition has a 2.1% fatality rate in the population. If you don't screen at all, 21 people in 1000 will die from lung cancer. If you screen, you end up saving about 10% of the people with lung cancer (about 3 out of that group of 21). But the screening means over 350 people end up being told they have lung cancer when they don't, 25 people end up being subject to an invasive procedure to diagnose the condition, and 3 (otherwise healthy but falsely diagnosed) people develop a major complication as a result of the invasive procedure.
Is all that harm done worth saving three lives?
The above is illustrated in this paper about lung cancer screening: https://t.co/YlbZIraekL
Also, if we had a cheap, 100% accurate test for “prostate cancer will cause you serious trouble in your lifetime”, i would guess it would be worth using it.
Or, better, that they lead to reduction in all cause mortality or reduction in days lost to disability?
I feel like it must be the first one otherwise people wouldn’t be arguing against it.
Thanks!
A hopefully not-too-inaccurate analogy might be if you heard a strange sound while driving your car, and took it into your mechanic, whose shop gets paid more the more tests they do and the more labor they put into fixing your potential non-problem, and they're potentially liable if anything later goes wrong that could be traced to that sound. Think about where the systemic incentives lie.
1.
At a societal level, perhaps false positives during screening put undue load on the medical system. But as an individual, it makes sense for me to screen for anything that does not take too high of an effort. If there is a false positive, it will be snuffed out in the next test.
2.
Here is a little thought experiment. Lets say the false positive rate is 1%, and you screen 100 people. Then, your total number of tests including repeated tests on false positive subjects will be ~101 on average (the chance of 2 or more false positives is low enough to ignore).
As long as the screening itself is not too much of an effort, then in fact false positives don't add that much load to the system.
So, we should be seeking additional methods for widespread screening rather than worrying about false positive rates.
---
I also find the doctors assumption of 10% false positive rates to be a little self serving in his argument. 1 in 10 people with an apple watch are going to be have atrial fibrillation detection? Doubt it.