Wednesday, May 26, 2010

Breaking 23andMe's Terms of Service: Not just the patient's problem.


A blogger over at 5Am Solutions Blog is about to break 23andMe's Terms of Service.

"So I called my primary care physician's office and told the appointment-taker I wanted to discuss my 23andMe results with my doctor. She said 'ok' and scheduled an appointment for next week."

May I just add. It is not the doctor breaking the Terms of Service here. It is the customer by bringing it in to their doctor.


"BOOM! That patient coerced that doctor into malpractice liability.Section 3 of 23andMe Terms of Service: “The Services Content is not to be used, and is not intended to be used, by you or any other person to diagnose, cure, treat, mitigate, or prevent a disease or other impairment or condition, or to ascertain your health.”

The worst of this is that 23andMe ACTIVELY INSTRUCTS its users to violate this clause —not only personally, but to also implicate their medical doctors in crime.

And the doctor is trapped: he can respect the law and alienate the patient, or ignore the law and appease the patient."

Oh, wait. Maybe by using 23andMe I am now involved in their legal mess? Crap!

That being said, I just received the Counsyl results from one of my patients yesterday. Unlike being put in a risky position by the good folks at 23andMe, Counsyl is straight up clinical and useful.

I will be notifying the patient via secure email of his results and spending an hour going over it with him.

The Sherpa Says: 23andMe, just like others in the space have demonstrated a general disrespect of the precarious position they have put physicians in by using such crazy and convoluted Terms of Service to avoid regulations. But heck, why should they care about the hot water they put us in.

26 comments:

John Kline said...

There is a dichotomy between genetics in the States and Europe. Issues of "Who owns my DNA"? crop up along with the fact that (in America) one pays a doctor, here one is assigned one free. That may not mean the doctor is any good, nor even speaks English well.

Andrew Evans said...

This is a bit of an overreaction, don't you think? From 23andMe's own FAQ section, under "Why can't 23andMe diagnose me?" - "If you learn that your personal genetic information suggests that you have a higher than average chance of developing a particular disease, you may wish to discuss your genetic information with your physician or another medical expert." While I think the Terms of Service clause is poorly worded, it seems clear that the intent is that this genetic information *alone* should not be used to diagnose a disease or assess health (which is good advice) and that a doctor with access to other information and expertise can use the information as part of a more comprehensive diagnosis. If that's not 23andMe's intent, then the FAQ wording on the public website calls the validity of the Terms of Service into question.

Steve Murphy MD said...

@Andrew Evans,

Doctors get sued every day by overzealous lawyers and patients. I would say no, it is not an over-reaction. It is based on the legally binding terms of service that the patient signed. But adding "anyone else" to the mix 23andMe is doing the over-reacting. They should drop that clause tomorrow.

Pharmacogenomic testing in and of itself is a diagnosis of poor metabolized state.

So too is BRCA testing that this company does.

This is the bullshit dichotomy here. They need to come out and follow Counsyl's lead with their clinical testing.

When they do, the ToS won't need to be so convoluted.

Andrew Evans said...

How is pharmacogenomic testing of metabolic variance a "diagnosis"? It's a test. It's not a diagnosis until it's interpreted by a clinician. Not all pharmacogenomic variation constitutes a disease state anyway - e.g., is increased response to beta blockers a pathology? Anyway, I think 23andMe's point is (or should be) that attempting to use the test result to self-diagnose (unless you're a physician with adequate genetics training - sadly, still a rare bird) is a mistake. I do agree that the wording in their Terms of Service needs to be changed, since it seems clear from their commentary elsewhere that they agree with what I'm saying in principle if not in deed.

Will FitzHugh said...

Hello Steve, thanks for pointing that out to me. Like many people, I probably don't read terms of services as completely as I should.

I went back and re-read it and I think you might be making too much of this. First of all, I only want to discuss my results with my doctor, not diagnose anything. I want to find out how much he knows about this kind of data and what his reaction is to it.

But even if it did want to diagnose something or modify a treatment plan (as in your Plavix example), it's not something I'd want my doctor to do solely based on my 23andMe results, anyway. I'd want him to use it as part of the picture and bring his medical knowledge and research into to it, too.

I agree that 23andMe's terms of service are convoluted, but they are that way because of the FDA's lack of clarity and slowness to come to terms with this kind of data. If 23andMe had waited to come out with a product with full FDA approval, they'd still be waiting. I prefer the current situation to that alternative. They are admittedly pushing the envelope and using some legalese to be careful about their liability, but I sympathize with their plight and don't see an easier path.

Steve Murphy MD said...

@Will,
What about my plight that if you don't like what I say or have a bad outcome you could use that ToS in a court against me?

The Plavix case is especially timely.

I wouldn't take people off Plavix who have no issues, Unless they had a 2C19 Poor Metabolizer status.

So I would be relying entirely on the 23andMe intepretation of their genotype data. OR I would just have you resend the lab to LabCorp for their exact same testing.

Steve Murphy MD said...

@Andrew,
Maybe you are mistaking genotype with result.

Allele *2 is an interpretation of a genotype.

So too is poor metabolizer.

the only thing that isn't this is reporting rs number.

Once you link the rs number TO AN INTERPRETATION like a link or like a disease or metabolic state it becomes a diagnosis.

The ideal service. Go buy your genotype at a rock bottom price somewhere. Ideally a CLIA approved facility. NYS facility if in NY.

Go get it analyzed somewhere else, like an FDA approved service.

Then be allowed to use it for diagnosis and treatment.

You arguing that the things 23andMe are testing for aren't medical diagnoses is disingenuous, even they won't be able to argue that one with the FDA or Congress.

Andrew Evans said...

I still don't see how this exposes the doctor to liability any more so than any other information a patient provides, though. I may be in violation of the terms of service (or not, depending on interpretation), but you're not a party to that agreement, and as long as you base your diagnosis on something other than 23andMe, why would you not be free and clear? You still need to follow your professional judgment and guidelines, with or without 23andMe results in hand. It seems to me that 23andMe's ToS actually protects you, because it isn't representing itself as a medical test, and therefore you're free to treat it like anything you hear from your patient. If your (untrained) patient says "doc, I think I have pheochromocytoma because I looked it up on the internet" - you don't order surgery, do you? It seems to me that the major implication of this is that physicians might be sometimes spurred to order secondary redundant tests, as you suggest, if it seems like the genetic markers are significant. I agree, it's stupid to run the same sample at the same lab just because of a poorly worded ToS - but this is more of my problem than yours, since me or my insurance will have to cover it - nobody's going to fault you legally for ordering the "official" test. As you've pointed out though, this is pretty much only relevant for pharmacogenomic markers (and I'll add Mendelian diseases), since everything else is pretty weak - if you're basing your diagnosis solely on weak signals elsewhere in the 23andMe results, you SHOULD be brought up on malpractice charges anyway...

Andrew Yates said...

"it seems clear that the intent is that this genetic information *alone* should not be used to diagnose a disease or assess health"

That's not what the 23andMe contract says.

Andrew Evans said...

@Andrew Yates -
But it's what 23andMe's public discourse says. Again, I agree that the ToS should be changed. But I think they would have difficult time defending a hardline position on it given their other public statements.

Beth Balkite said...

So you can't just discuss something with your doctor because the doctor did not order it? Sounds more like control than good patient care.

Steve Murphy MD said...

It's not about discussing. It is about factoring that information provided into patient care and diagnosis. It is almost as if you have to reorder the same test, only this time it has to be from a lab that has the stones to say it is a clinical test.

Sounds crazy right? It is and it is why the ToS for these DTCg companies needs to change.

Dan Vorhaus said...

@Andrew Evans is correct. Steve, the ToS aren't binding on you. You do with the 23andMe data - and any other piece of information provided by one of your patients - what your professional judgment dictates. There's no ambiguity there.

Andrew Yates said...

@Beth Balkite

"So you can't just discuss something with your doctor because the doctor did not order it?"

No, you can't just discuss 23andMe as you choose because the 23andMe Terms of Service contract bind you to certain restrictions including clinical application which would include medical consultation with your doctor.

Your hostility is misplaced. Blame 23andMe's lawyers, not your doctor.

Andrew Yate said...

@Dan Vorhaus

http://www.thinkgene.com/dan-vorhaus-attorney-at-law-legally-advises-medical-doctors-to-use-23andme-while-providing-medical-advice/

"Dan Vorhaus, Attorney At Law, Legally Advises Medical Doctors Can Use 23andMe To Provide Medical Advice"

Keith Grimaldi said...

What do the ToS really mean? 23andme say we will genotype your DNA but you are not allowed to use the results for health purposes. If I go and do so, I am breaking my agreement with 23andme - so they can sue me? Of course they wouldn't, I presume.

But is that it? 23andme tell me not do do something, I disobey them and take them to my doctor. I'm not committing any crime or breaking any medical ethical barriers am I? Nor is the medic.

The medic might need to exercise some judgement though as to whether a particular result (e.g. for something relatively serious like drug metabolism or breast cancer risk) should be repeated/confirmed by another lab. On the other hand advice to reduce asparagus consumption to avoid smelly pee may not require a confirmation.

So is there really a legal issue here? I know it's a nuisance for medics, all these patients coming in with results, information, questions - but that's the internet in general, not just genetics.

Steve Murphy MD said...

@Keith,
Because of 23andMe's terms of service I can have ZERO confidence that the test can be used for any medical judgement. Why? They prohibit its use by the participant OR ANYONE ELSE for medical decision making.

If I use it to manage a patient, I am using a test whose makers say shouldn't be done. In fact they believe that so strongly they put it in the Terms of Service.

So when I am on the stand as a patient I told Plavix would work for them when instead they had another heart attack for ANOTHER reason, and I am asked to explain why I didn't order the clinically valid test, what should I say?

A genomicslawyer with a specialty of ELSI rather than malpractice law told me it was ok?

Or maybe, Dr Keith told me it was ok? What about 23andMe who told me it was ok? Oh wait, they didn't say that.....

In fact, their Terms of Service prohibits that.

They say it's not clinical, but then tell you to take it to a doctor....

Stupid. They need to change their ToS and accept responsibility.

Steve Murphy MD said...

@Dan
No offense, but my attorney specializing in healthcare law has advised us not to use the 23andMe product BECAUSE OF its Terms of Service.

So, I will not use any test which has that in their terms of service. And since these tests will soon be ALA Illumina style Rx, they will need to change, not me.

Andrew Evans said...

@Steve -
Indeed, you shouldn't base your whole diagnosis on a test that doesn't represent itself as a medical test. You shouldn't base your diagnosis on Wikipedia articles either. Or the fact that your patient tells you he's *sure* he's got a rare cancer because it came to him in a dream. As a professional, you need to follow up the way your judgment sees fit. That's why you get paid the big bucks. While it's a shame that 23andMe took your patient's (not your) money for a test that is inadmissible, you can weigh what your patient tells you and decide if *all of the other evidence* merits duplicating the result.

You are focusing on whether you'd have to take the stand in a case where you allowed yourself to be lulled into accepting Plavix based on 23andMe alone (and the patient had another heart attack), so let's look at that for a moment - why would you ever do such a thing, professionally? Even if the 23andMe test had a different ToS, would take any test results that a patient hands you out of the blue, that you didn't order (or another doctor didn't order) and not follow up on them yourself? If this were a boring blood test we were talking about, and not genotyping, would you do it then? I sure hope your answer would be "hell no..."

So - while I agree, as a 23andMe customer, that it sure would be nice if the ToS made more sense, there still isn't really any increased liability here for you as a doctor that I can see. You're not a party to the ToS, and you're not absolved from basic professionalism any more so than you would be with any other patient-provided information. So aside from all of us agreeing that 23andMe should man up so it can save patients and insurers the cost of redundant tests, this is a tempest in a teapot.

Anonymous said...

Every patient/clinician encounter is unique. That said, a policy for the unlikely event of a 23andme customer (how many customers are there only about 30-40K last time I heard a number)seems relatively simple to implement. If a patient comes in and says that by using a drugstore high blood pressure tester they have found that they have high blood pressure, how do you proceed? Seems like an analogous situation. Is there a real actionable health problem or potential health problem? If so, you take whatever further action you deem appropriate as a licensed practitioner in your specialty... tests you order to confirm then interpretation and action from you, referral to medical geneticist or other specialty etc. Your possible actions do not depend on 23andme information; this is merely the motivator for the patient. 23andme patients are likely interested in preventive medicine, lifestyle advice is same no matter the genotype in most instances. Finally, if patient doesn't operate within your groundrules can't you "fire" them?

Steve Murphy MD said...

@anonymous drug store bp cuff

No, this is not the same. Not even close.

The BP cuff doesn't make you sign a terms of service preventing you from using that information to make medical decisions.

Steve Murphy MD said...

@Andrew Evans

"Indeed, you shouldn't base your whole diagnosis on a test that doesn't represent itself as a medical test. You shouldn't base your diagnosis on Wikipedia articles either. Or the fact that your patient tells you he's *sure* he's got a rare cancer because it came to him in a dream."


I just love how you equate a biochemical analysis of human biomedical sample with "Dream Visions" and Wikipedia articles!

That being said, would I repeat a BRCA test that someone told me they had? No I would call the lab and ask for validation or the report.

Would I do the same for a 2C19 result from 23andMe? The answer is no. I think we all know the answer here. No doctor in their right mind will not Re-Order the 2C19 test.

Thus this whole adventure is a waste of the customers money, because some company wanted to create a bad faith contract and avoid any and all liability.

Nice.

Andrew Evans said...

@Steve -
"I just love how you equate a biochemical analysis of human biomedical sample with "Dream Visions" and Wikipedia articles!"

I think we both can agree that until 23andMe changes their attitude, from a clinical perspective, they might as well be testing unicorn blood! :)

"Thus this whole adventure is a waste of the customers money, because some company wanted to create a bad faith contract and avoid any and all liability."

We're in complete agreement there as well...

Anonymous said...

Dr. Murphy,

Thanks for a good post, one of many. You don't buy the bp cuff analogy. Fair enough.

That does not alter the point that a 23andme policy should not be hard to implement for the "average" internal medicine practice in which a 23andme user is a low frequency event.

In a specialized practice with a higher frequency of 23andme account holders, conservative legal advice does seem appropriate given the current 23andme TOS.

IMHO, in either case, staff should not be writing down that the reason for a patient's visit is driven by a particular commercial entity.

A doctor's "TOS" is in the physician's hands, so not discussing info obtained from a given consumer genetics company but interpreting info from another is a practice that a patient would have to live with.

Steve Murphy MD said...

@Andrew Evans and @Anonymous,
You should both come to CGC in Boston. I will be addressing this in my talk.

Anonymous said...

A result such as poor metabolizer is not a diagnosis. It's a fact. It's a data point. The problem here is the cookie cutter method MD's are taught in the curriculum used by med schools. They can't understand anything they actually have to think through - someone is always doing their thinking for them.

There is a sea of information. The intelligent person learns how to ride the waves. That includes the patient if no doctor is available by choice or chance.