The Public Health Case for Direct-to-Consumer Personal Genomics

OK - A couple more thoughts on this move by health departments in California and New York to regulate personal genomics. I’ve made my quasi-libertarian case that this is my information and shouldn’t be mediated by an under-informed (and possibly antagonistic) physician gatekeeper. And I’ll leave the companies to make their own case on the issue of lab oversight.

But now let me make an argument on public health grounds - the home turf, after all, of these state agencies. To my mind, their actions will directly contravene their own mandate, and will have the result of reducing the public’s health.

The California DPH says it’s acting to “protect consumers.” As Wired Science’s Alexis Madrigal ferreted out, in a nifty bit of reporting, CDPH’s Karen Nickel said in a June 13 meeting that the state’s primary concern is that personal genomics companies are creating the “worried well” - citizens who stumbled into a level of knowledge about their genome that they were unprepared for, and now have may be fretting in a a way detrimental to their health & well being. Put aside the fact that, as a public health matter, the “worried well” is a supremely thin basis for action (what pray tell is the prevelence of “worried well” in California? The incidence? The relative risk of learning one’s genome? What sort of epidemiological studies have been performed to measure this population?). And put aside the fact that, as others have noted, the customers of 23andMe and Navigenics and other personal genomics companies are, in demographic terms, probably the *least* likely to be categorized as “uninformed” or naive. These are early adopters, they’re paying lots of money (opting-in), and are probably far more prepared to reckon with genomic information than the typical citizen. But put that all aside.

My argument is simply that by restricting personal genomics to a physician-vetted service, these state public health departments would be eviscerating the actual public-health utility from genomics. The whole *point* of learning one’s genomic predispositions is as a predictive and preventative tool. Learn early, so as to change our behaviors, intervene early, and either skirt or reduce the prospects of disease. This is a *long term* tool. But by regulating the service, these state health departments would severely impinge the opportunity to make the largest public health impact, in two specific regards:

1) Public health, by definition, is about populations, not individuals, and Nickels makes a quasi-population argument when she identifies this group of “worried well.” OK, let’s take a stab at quantifying this. The worried well would be some fraction of personal-genomics customers; let’s give Nickels a big gimme and say that 20% of customers would somehow overreact in a way that’s detrimental to their health (I’m of course making that up, and it’s almost absurdly high, but let’s go with it). Stress would be the most obvious detriment, but it could be something like taking unnecessary medication or supplements, etc.

Now consider what percentage of personal-genomics customers actually engage in their genomic information in a way that’s *beneficial* to their health, as intended. These people pay their money, get their results, spot their risks, and change their lives, often in small ways or rarely in big ways. Let’s low-ball it and say that 60% of all customers act on their results (paying $1,000 or $2,500 is actually a significant motivator, but let’s assume 40% just ignore the results altogether). And of course not all results would be positive; some would be null. So let’s take another slice and say some fraction of the 60% actually measurably benefit, either in peace of mind or in some slight amelioration of diet and exercise or doctor’s visits. Let’s say half of the 60% - so 30%. Even at that low number, that is very clearly a net positive - the public’s health at large has been improved.

Of course, I’m making up these figures, and really it’s probably impossible to measure (though I bet 23andMe and Navigenics are crafting customer surveys to help fill in this picture). But really, by any informed assessment, the net potential for improving the public’s health far outweighs the possible detrimental effects of “worried well.” Just as vaccines and even exercise already have their detrimental side effects, sure, so will personal genomics. But if you’re tasked with improving the public’s health, as these agencies are, why not consider the benefits as well as the risks?

2) So genomics are useful as a predictive tool, they give us a peek into our longterm health prospects and an opportunity to intervene and improve those prospects. The fact that consumers in California can, for the moment, engage in that information now at their own behest means they are getting the information when they want it, which is by definition as early as possible.

So what’s the logical consequence of forcing a physician into the picture as a middleman? Well, it’s a pretty good guess that it’ll delay people from getting the information. Put physician-phobias and reluctance to schedule a visit and all sorts of other procrastinations together, and I think this would result in less and later genotyping; both a significant delay in when this information reaches people - as well as a significant reduction in the number of people who actually bother to jump through these extra hoops. The net result, again: a squandered opportunity to impact the public’s health.

I would be the first to acknowledge that the actual science is fairly raw here; we’re in the early days of using our genomes for actual health decisions. But that’s the point: better to get familiar with the information now, when it’s fairly low-impact, and work out the kinks than to wait for the science to somehow emerge fully-formed and neatly packaged. Because if we’re waiting on the physician community for that day, it’ll never come.

But assuming the public health department acknowedges that genetics *does* have some utility for our health, then I’d remind them that a fundamental principle of public health is awareness - give citizens information earlier so that they can avoid putting themselves at risk. That principle drives public health’s actions against smoking, infectious disease, sexually transmitted disease, natural disasters, and so many other threats. Likewise, it drives their actions on positive behaviors like proper nutrition and exercise. So why, in the case of genomics, should the same principle not apply? Why, in this case, do state health departments think the public should be prevented from learning about their risks?

Published by: tgoetz on June 27th, 2008 | Filed under Epidemiology, Technology, Policy, Genetics

12 Responses to “The Public Health Case for Direct-to-Consumer Personal Genomics”

  1. Steven Murphy MD Says:

    This is a great argument. But, it is predicated on 3 fallacies.

    1. The testing indicates risks that are modifiable by current interventions. (not true for 9p21.3 or fgfr2, I could name about 30 others) Vaccines have way more proof of benefit than these tests

    Your words “So genomics are useful as a predictive tool, they give us a peek into our long term health prospects and an opportunity to intervene and improve those prospects.” Once again, not always true

    2. That a delay in testing is actually a bad thing. I would argue, a delay allows us to gain more insight into the real risk prediction of these tests.

    3. Physicians will never be prepared to handle this……I call this just as arrogant if not worse than me saying the public who did not take biochemistry, physiology, cell biology, histology, molecular biology, genetics (in undergrad), will not be able to understand….Thomas…..come on now….

    3 fallacies….that is why your argument doesn’t hold water with me or the government….

    Just my 2 cents


  2. Steven Murphy MD Says:

    Oh and because the data they receive is less than accurate at best.

  3. tgoetz Says:

    Thanks for the comment, Steve. Just to quickly clarify:

    re: #1: I’m not saying every risk is modifiable. That’s obviously not true. But most diseases are not entirely genetic, they reflect the interplay of genes + environment.

    #2: obviously debatable.

    #3: I think you read me wrong: I’m not saying physicians will never be prepared to handle this. My sentence refers to waiting on physicians for a “fully formed and neatly packaged” genomics tool that will make sense to consumers. We can’t wait on physicians for *that* because it’s not their job.

  4. Valentin Says:

    > Why, in this case, do state health departments think
    > the public should be prevented from learning about their risks?

    well, these health departments are probably mostly made up from doctors and these are simply protecting their home turf; their comfortable market. May sound cynical, but I have some insights into health politics (although not in America) and I’m quite sure that’s really a big part of it.

    And btw. in my browser at least (FF3, Windows XP), the font in the comment- field is ridiculously small; unreadable without using the ‘zooming’ function.

  5. Joe Six Pack Says:

    These DTC companies are mostly presmised upon a entirely flawed (stupid?)
    business model. Irrespective of what regulators do, the majority -
    perhaps all - are likely to disappear. The libertarian in me believes
    the small number of fools who would pay exhoribitant sums for essentially
    worthless information should be allowed to do so. However, one does
    hate to watch others get conned.

  6. Mr. Gunn Says:

    Mr. Gene Sherpa is at it again. It doesn’t matter if the information is actionable. People should be able to know what’s in the DNA simply because it’s their DNA. There’s lots of info revealed by SNPs that has nothing to do with medicine, such as ancestry and taste preference and so on. Why should doctors be the gatekeepers of such non-medical information? Since everyone agrees that doctors aren’t equipped to handle this, why would you want to require a referral from them that they’ll never want to give? Finally, whole idea of collecting SNP data sets from as many people as possible is to improve the predictive power of the analysis. Stopping the collection will hurt this, so there’s no way that we’ll get a better idea of the relative risk indicated by SNPs by not collecting more data. It’s precisely the opposite.

    I appreciate that Steve is now actually addressing points, instead of just chanting “SNP analysis is medicine”, but his arguments are so clearly and transparently wrong that it continues to amaze me that he even bothers. Steve, you’re not doing your quest to be hired as a consultant by these companies any favors by making your lack of understanding part of the public record.

    Joe Six Pack(possible Gene Sherpa sockpuppet?) - You’re just offering opinion, knowing nothing about the market forces driving these services ever cheaper. Do you really believe that no useful information will come out of a large collection of SNPs across race, gender, and age?

  7. Joe Six Pack Says:

    I do not believe any useful information will come
    from SNP tests sold to the general public. In
    order to give utility to the information derived
    from these tests the assays must be performed,
    and the results studied, in settings in which
    investigators can attempt to control for variables.
    Once a potential correlation is found, it must be
    prospectively validated for it to have proven meaning.
    Most of the DTC companies have no ability to obtain
    accurate clinical information from the very few
    customers they are likely to have, and could not
    be trusted to conduct what would prove to be
    extremely expensive studies if they did.
    Finally, relatively small increases in relative
    risk for unlikely events have no meaning unless
    placed within the context of the totality of
    risk. If, for example, sucking on cancer sticks
    increases my relative risk of getting lung
    cancer 30 times, and lung cancer is rare
    among non-cancer-stick suckers, of what use is
    it to know that I carry a SNP that is
    associated with a 2-fold increased risk of the
    disease. This is especially true since I may
    carry as yet unknown SNPs that are protective
    to an equivalent extent.

  8. Steven Murphy MD Says:

    Mr. Gunn,
    Hyperbole will getcha everytime….

    “There’s lots of info revealed by SNPs that has nothing to do with medicine, such as ancestry and taste preference and so on.”

    Really? That’s news to me….we still don’t know what most of these means…even non-medically. I just spent 2 hours discussing with several Yale PhD’s what some of this gobeldeegook means.

    “Why should doctors be the gatekeepers of such non-medical information?”

    No one is saying they should be….provided it truly is NON-Medical……..

    “I appreciate that Steve is now actually addressing points, instead of just chanting “SNP analysis is medicine”

    I said “SNPs+ Analysis=Medicine” in the context of medically useful SNPs, Mr. Gunn

    “Steve, you’re not doing your quest to be hired as a consultant by these companies any favors by making your lack of understanding part of the public record.”

    Mr. Gunn, I sincerely regret to inform you that I am an MD who developed a genetics education curriculum for physicians, a fellow in Clinical Genetics at Yale, and sit on the Informed Cohort Oversight Board of the Coriell Personalized Medicine Collaborative with Charles Rotimi I didn’t stumble upon these. I was invited because of my insight into the entire field and topics precisely related to what we are talking about now…….I understand if you were not aware of this… harm….no foul

    “Joe Six Pack(possible Gene Sherpa sockpuppet?) - You’re just offering opinion, knowing nothing about the market forces driving these services ever cheaper. Do you really believe that no useful information will come out of a large collection of SNPs across race, gender, and age?”
    Shame again……do you know who Joe is???? What Joe’s qualifications are????
    You know what they say when you ASSume…..

  9. ADAM HANFT Says:


  10. Mr. Gunn Says:

    Like I said somewhere on the twenty or so sites we’ve sparred on, I’m asking these questions in an attempt to make sure that everyone has a well thought out position that incorporates past history but also includes viewpoints from people who aren’t “in the trenches” and aren’t as likely to miss the forest for the trees.

    I don’t mean to be so rhetorically aggressive, but it’s the only way I can get an real response from you. I don’t have your decades of experience in genetic testing; I’m a smart guy who can reason as well as anyone, and what I’m seeing just doesn’t add up. You say that the fact that there’s non-medical information revealed by SNPs is news to you, but then claim two quotes down that you meant your comments only in the context of medically related useful SNPs. Well, which is it? Did you know of non-medical uses while you were pushing your “only on doctor’s orders” position, or is the existence of non-medically related SNPs news to you?

    Again, I’m asking this to make sure that the judgement of people with experience isn’t clouded by a failure to see something truly new for what it is. SNP profiling is a totally new and disruptive technology, much like file sharing is disruptive to recording companies and the internet is disruptive to traditional media. We can and should learn from these examples. Much like the internet was in the 90s, SNP profiling is a technology whose potential we can’t properly appreciate just yet, but one thing we do know is that we’re right at the point where we have massive computing power and sharp analytical techniques, and the only thing missing is the massive data sets for them to work on. Health care is in crisis! Something has to be done to reduce costs, and I’ve not seen anything with as much potential to do so as SNP profiling practiced on a worldwide scale, but you’re still worried that someone might get told they have a 25% risk when it’s actually 20%. I understand all about the tradition of being a curmudgeon regarding new technology because it never lives up to the hype. I wish we had some sort of crystal ball that would allow us to figure out exactly the hype level we need, but lacking that balance is achieved by young people like me promoting that in which they see promise and older more experienced people such as yourself trying to burst our bubbles.

    With that in mind, let’s put aside out differences and see if we can come to some sort of experienced veteran + young idealist consensus vision here. We both agree that there’s a need for regulation, including CLIA certification and use of validated methods approved by geneticists on staff. Can we agree that a doctor’s order prior to using the service isn’t necessary, since a concerned individual can always get a referral to a geneticist after getting their information?

    JoeSixPack - The only way to solve those problems you rightly point out is to collect more data. I don’t agree that the companies have no incentive to perform the studies, because there’s a ton of interest in academia for doing these studies, and I find it hard to believe a company connected so strongly to Google and in that environment won’t want to collaborate. At any rate, in a few years the cost of fully sequencing an individuals genome will drop dramatically, and if private enterprise doesn’t do the research, a public consortium will. At that point we can finally address your good questions, and the alternative is what? Do nothing? Carry on practicing medicine like we always have and not see if there’s at least something we can learn from this? Almost certainly this won’t live up to the hype you may be hearing in some circles, but you just can’t stop certain people from hyping things, unfortunately. That doesn’t mean the technology doesn’t have great promise.

    If we don’t do it here, someone else will, and can we afford to lose the lead here? Can we afford to piss away a hundred billion dollar industry, one that actually can improve health instead of compromise it like most industries, because people might get worried when they find out they might have a risk factor?

  11. David Hamilton Says:

    Right off the bat, I agree with Tom that there’s not much utility at this point in forcing M.D.s into the personal-genomics equation. The unfortunate fact is that most doctors are on the trailing edge of medical innovation, which is why it typically takes 17 years or whatever for confirmed medical findings — like treating ulcers with antibiotics — to find their way into general practice.

    That said, I’m a lot more skeptical about Tom’s argument that genomic results are going to lead people to take beneficial actions for their health. First, the most beneficial actions most people can take are to eat right, exercise more and stop smoking. No one needs to pay $999 or $2500 to learn that, and while some people may be more motivated by the expense to actually change their behavior after paying for genomic tests, the history of behavioral change motivated by statistics (as opposed to, say, the new social stigma attached to smoking) suggests otherwise.

    Second, where people are made newly aware of medical risk, odds are good that they’re going to demand whatever quick fixes they can find. To take just one example, someone learning about a 2x increased risk of lung cancer is probably more likely to “talk to their doctor” about something like regular CT scanning, particularly if it’s being heavily advertised by hospitals or medical centers in their area. The fact that the diagnostic technology is itself of limited demonstrated benefit won’t matter, and of course purveyors of what is a pretty lucrative service will undoubtedly market to personal-genomics customers as soon as the services become sufficiently common.

    Third, it’s entirely possible that growing demand for unneeded or marginal services could actually harm public health. CT scanning involves a lot of radiation, so personal-genomics results might easily lead people to increase the risks to their health instead of reducing them.

    So while I don’t know the incidence or prevalence of the “worried well,” we certainly know they exist, and while they may not be the largest driver of overall healthcare utilization and rising costs, their impact is almost certainly nontrivial, and so can’t readily be dismissed out of hand.

    A final point on the overall societal benefit of personal-genomics “data collection.” The main problem is that these datasets are likely to be of limited utility, since any association studies must rely on the self-reported phenotypes of personal-genomics customers. So it’s going to be very, very noisy data at best. Reported associations are already tentative enough without introducing additional sources of error.

    I’ve written more about the subject over at BNET, and will probably take another crack at the subject before long.

  12. Paola Says:


    Thank you for sharing your views on the persona genomics issue. I would like to agree, however, with those who provide very good arguments against your position.
    You are right when you say that not all risks are modifiable, and that there is an environmental component to almost all health conditions. If that is the case, and the latter IS modifiable, why learning about genetic traits of unknown significance in the first place? People do not need to pay thousands of dollars to finally realize that living a healthy life is actually a good thing!

Leave a Comment

You must be logged in to post a comment.