Why Normal Matter More Than Sick

The new May issue of Wired (the "Get Smarter" issue) has an essay I wrote about why medical science should spend more time studying the baseline - i.e., what's "normal" - and less on the aberration.

Here's the gist:

For all sorts of conditions, there's often no definition of normal. In heart disease, for example, CT screening tests can spot abnormalities in arterial plaque — but no research exists on whether that information is actually predictive of heart disease or stroke. "We need to know normal variation," says Pat Brown, a professor of biochemistry at Stanford University School of Medicine. "It's really underappreciated as a part of science."

This essay came out of several conversations with Brown, an oncologist at Stanford who's one of those scientists whose knowledge seems to span into every possible corner - and can articulate relevance in clear, compelling terms. He has been ranting (in the best possible way) about the need to know normal for a while, and when I couldn't get him to write this essay himself, I did the next best thing: pulling as much of the idea out of his brain as I could muster.

And it's something that many scientists, particularly those working at the molecular and DNA level, are saying: We need to know more about the baseline. I spent the past couple days at the Institute for Systems Biology symposium (where Bill Gates gave the closing keynote), and many of the presentations touched on this theme: there's too much variation at the cellular level to be able to clearly say (in the plainest terms) this cell is bad and this one is good. Only when we understand the variations at a cellular or molecular level will we know when to intervene on disease, or whether treatments are working, and so forth. As the story makes clear, this is especially important for cancer right now - are we overtreating or undertreating? - but it will become a relevant issue for all intervention soon enough.

And push it forward: Should we know what's normal not just in the human population, but in our own body? For instance, a fellow from the ISB told me about an experiment that the institute conducted one day: They had several staff in and took blood samples, then went out and had lunch. After the meal, they took another blood draw. Sure enough, the before/after blood makeup was totally different - not just in terms of stuff like glucose levels but down to the protein level. Long/short: even in one person, it's difficult to know what's "normal" on a protein level. But the more we can discern these baselines, the more we'll know when to treat, and how, and even whether.

More on the symposium later, if I can manage.

Enter Navigenics, Where Personal Genomics Gets More Medicial

Today marks another entrant in the personal genomics game: Navigenics, the much anticipated startup out of Redwood Shores, Ca, is open for business.

The company arrives as direct competition to 23andMe and DeCodeMe, both of which began offering direct-to-consumer genotyping last year. Navigenics was originally planning to launch around the same time as the competition, but ended up taking several months longer to fine-tune it’s product. As planned, Navigenics is taking a more clinical approach to personal genomics, with a more overt pitch towards the medical implications.

I had the opportunity to visit the company last week and get a preview of the service. Here are a few standout observations.

1) The Results: Navigenics launches offering results on 18 diseases, from glaucoma to colon cancer to Alzheimer’s. This is about the same number as 23andMe had at it’s launch, though they’re now up to 58 different conditions.

One big distinction is that 23andMe lets users peruse the entire results of their genotype run – more than 500,000 different SNPs. Navigenics, even though they’re using a 1 million SNP chip (as does DecodeMe), is more circumspect with its results, only letting customers see the results for those conditions they’ve vetted.

2) The Business Model: As has been anticipated, Navigenics will charge an initial fee of $2,500 for a one year membership – and then an annual fee of $250. This compares with about $1000 for permanent access at 23andMe and DecodeMe.

That's been criticized as a bad deal, especially since you can't look at the 1 million results. But Navigenics offers an intriguing twist: It will freeze your spit sample, allowing the company to re-test your DNA as more associations with different SNPs are discovered (and deemed scientifically valid). Mari Baker, Navigenics' CEO, says they expect to go back two or three times a year to extract more data points.

That means that there's a clear trade off: with 23andMe, you're buying into today's technology, and they promise to show you everything they have. With Navigenics, they're not going to show you everything, but they promise to keep you up to date as the technology and the science improve.

3) The Calls: One thing you notice when you get your 23andMe results is how subtle the differences are between the average person’s risk for disease and your own. For colorectal cancer, for instance, my 23andMe results tell me that I have a .21 out of 100 chance of developing the disease, compared to a .26 out of 100 average risk. That may be scientifically valid distinction, but as a consumer it is so slight as to be no difference.

Navigenics uses a different method of calculating your genetic risk. I won’t get into the details here, but basically they make a “Lifetime Risk analysis” that results in what they believe are stronger calls. Certainly the numbers are more emphatic; examples I saw showed, for instance, a person with a 51 percent risk of heart disease, compared to an average 42 percent risk. That’s a striking difference.

4) The Physician: Navigenics puts a great deal of emphasis on the utility of genotype data for useful medical insights. It’s clearly one of their main selling points. To that end, “we’re putting education towards the top of our agenda,” says Mari Baker, and they’ve bankrolled an online continuing medical education course on Genomic and Personalized Medicine with Medscape. What’s more, they suggest customers bring doctors their Health Compass Report, a primer for personal physicians explaining what the company does, how it calculates risk, and what their patients results might precondition them for. To me, this is a bold and aggressive endorsement of the power of genomic data for real-time medical insights, much stronger than anything 23andMe has done. It’ll be interesting to see how the medical community responds.

So it’s a more high-fidelity environment, with less flexiblity for the user. And though 23andMe gets lots of attention because of their founders’ Google connection, make no mistake: Navigenics is as blue-chip as they come, with top management from T-Gen and Kleiner Perkins.

Long and short: Yesterday personal genomics was an oddity. Today, it’s an industry.

This is a cross-post from WIRED SCIENCE

Practicing Patients: The Response

Thought I'd jot down some of the feedback I've been getting to my story in the NYT Mag on PatientsLikeMe. Overall, I've been fairly blown away by the response – I've gotten dozens of emails and the story has been been blogged mightily. In stories like this, where I'm writing about one company or person, it's important to keep in mind that the enthusiasm is more for the ideas and portent of PatientsLikeMe and not my story, per se. But I'm going to assume that the humble messenger - me - did a fair job conveying the import of the message, and that that's worth something. Anyway, the reactions fall along a few lines:

1) Patients are thrilled. I've gotten some striking messages from people with a chronic disease or from people related to a person with disease, for whom a resource like PatientsLikeMe is a godsend. Many patients with chronic disease are, surprisingly, simply left confused by their interactions with the medical world - doctors and specialists all telling them slightly different things, giving them slightly different courses of action and treatments. The ability to connect with others like themselves and then, moreover, to take *their* advice on treatments seems remarkably clarifying for these patients. " went to a conference in Tampa," one wrote me, "and different doctors believe in different treatments with different patients. Is this because of where they were trained or evidence-based medicine?"

It's a fascinating problem - one I hadn't anticipated. I knew patients would be eager to get resources other than their doctors; I knew that they were frustrated with the time they get from their doctors; but I didn't realize that they were so frustrated with the *information* they're getting from their doctors.

2) Doctors are leery. I knew this was out there, and I tried to make sure the piece reflected the skepticism, and in some cases antagonism that physicians might have towards a resource like PatientsLikeMe. This isn't just for competitive reasons (though there is that element). I think doctors are legitimately concerned about their physicans patients embarking on treatments that may hurt them rather than help them, and that the power of PatientsLikeMe's data - it all looks so convincing - might compell some patients to disregard advice or treatment to their detriment. But another aspect to this is that doctors aren't comfortable, themselves, with the idea of their world being transduced into data.

Medicine, from a physician's POV, is still largely empirical. Diagnosis and treatment is often guesswork - educated guesswork, but still guesswork. They are trained in science - educated in science - but that doesn't mean doctors are all comfortable with the idea of data-driven decision making. (read Jerome Groopman's How Doctors Think for a startling - to my eyes - reminder of how much variability there is in physicians' diagnosis). Since PatientsLikeMe reduces all this to data - treatments and results are right there to be correlated - it's an unnerving glimpse into the future.

3) The field is wide open. Among general readers, many have fastened onto the story's penultimate paragraph:

Really, when you start looking, information can be found everywhere. If we could gather in structured communities and create databanks to inform our approach to life decisions, not just health decisions but also gardening or parenting or car-buying decisions, we could do everything in a more informed manner. Were we all to avow a philosophy of openness and churn our experiences into hard numbers, we could presumably improve our odds in all sorts of decisions. Why not a PregnantLikeMe or a ParentsLikeMe or even, really, an all-encompassing PeopleLikeMe?

First off, as a writer it's terribly gratifying to know that folks actually made it to the end of a 5,000+ word article. So that's nice. But really, the idea here is one that I've been mulling over for a few months - the power of data to be collected and deployed in unanticipated places for unexpected results. I'm especially interested, obviously, in the potential of data to affect health-care, both in terms of predicting and averting illness as well as to treat it (more - lots more - on this later). So it's doubly gratifying that this idea has such resonance among readers - that ordinary folks get this idea, that there is power in ordinary life, in unharnessed information, to guide our actions. It's the idea, really, that some sort of meta-analysis of daily life could go on (should go on?) that would let us *really* learn from our mistakes. Obviously, as the next graph in the story made clear, I'm not the first one to have this realization. So it'll be fascinating to explore where this idea goes, and how innovators may respond to both take advantage of the opportunities for data collection and aggregation as well as to respond to the demand individuals have to make better decisions.

4) Some metrics. So since this was such a data-intensive story, I thought some metrics might be illuminating.

7 6 - highest rank of story on NYTimes' Most Emailed list. [corrected! Someone caught it at #6, & passed along the JPEG to prove it]

30 percent - Amount enrollment at PatientsLikeMe has increased in seven days.

10 - factor by which enrollment at PatientsLikeMe mood community increased (from about 100 members to somewhere just over 1,000) in that time.

14 - Number of companies who've written to tell me they loved the story - and wouldn't I like to speak to their CEO? (Honestly, I appreciate these emails & have learned of some interesting endeavors because of them)

The WHO's Silly Embargoes

I've always hated the presumptive embargo - the company or academic journal or quasi-governmental agency that sends out an email blast of "news", but stamps the word "EMBARGO" onto it, stipulating that the information cannot be published until a certain date. Not only does this practice undermine the concept of "news", but it often presumes that the journalist will play ball for no reason other than to get another such email later. As a matter of practice, I disregard the defacto embargo, and consider any press-release sent to me by email fair game, unless I have previously agreed to abide by a hold. That said, I will also admit that never, not once, have I been sent something on presumptive-embargo that I consider worthwhile, so I've never had the occasion to test this principle. Anyway, that's a long preamble for another tale from the World Health Organization's silly use of embargoes and penalties, in this case removing the Associated Press from its distribution list because they violated an embargo and ran a story announcing the eradication of polio in Somalia. To be clear, my sourcing on the WHO penalizing the AP comes from this blog post.

Now the AP and other media like the New York Times - which broke a WHO embargo last year and was likewise banned for two weeks - actually do agree to abide by these embargoes on an ongoing basis, in order to be on the distribution list. But this system reeks to me of an antiquated and impossible-to-maintain closed world of information that fails to serve anybody's interest, not the WHO's nor the media nor the people of Somalia or whereever the WHO may have waged a success. The WHO clings to the embargo because it's a way to maximize the exposure of their release - let everyone publish at the same time, starting at a set time. It's the same principle academic journals use in promoting their "breakthrough" studies. But just as these staid journals are being challenged by the power of open information via open access publishing, so the WHO - of all agencies, after all - should realize that by losing the charade of 'exclusivity' they might gain greater coverage. Put out your press release, but don't stick to an artificial charade. Let the AP dash off their report - that's their value as a news source. Let the NYTimes dwell on the news, add context and expertise, and publish their more thorough story - that's *their* value. And so forth, on to blogs, like this one, that may in fact be interested in the news but aren't privy or won't cotton to such embargoes. Who knows? Maybe we'd be writing about polio eradication, too, instead of the WHO's stupid short-sighted media policies.

My Latest Story: PatientsLikeMe in NYT Magazine

Tomorrow's New York Times Magazine has a story I wrote on PatientsLikeMe, the innovative patient community. The company, I believe, is doing some powerful things:

1) It's making a bold call for Openness with our medical and health information, arguing that this information is more powerful out and shared and aggregated than locked up and protected.

2) It is an emerging research tool, allowing patients to gather together and consolidate their sometimes anecdotal subjective experience with disease into collective bodies of knowledge, that can create real-time insights into treatments.

3) It is, perhaps most of all, a way for individual patients to actively improve the way their manage their disease. My feeling, as I've written here before, is that all Americans are going to fall under some rubric of 'diseased', in part simply because we're diagnosing earlier and identifying risk earlier. So we will all be managing our health actively, by the numbers, as the members of PatientsLikeMe are now.

If you read the story, please drop me a line to let me know what you think - or better yet, comment here, so that we can craft a dialogue. I'll be following up with reactions over the next few days.

Oh, and here's a link to an interview I did on the Brian Lehrer Show on WNYC about the story and some related issues concerning off-label prescriptions.

Back to Basics: the Human Brain

Funny how when titans of technology, guys who made their millions (or billions) as profound innovators, turn to science, they often just want to do the simple stuff. Bill Gates, for instance, has geared his foundation to accomplishing simple goals (albeit on a grand, transformative scale), and now word that his fellow Microsoft founder Paul Allen - shown here - has turned the Allen Institute for Brain Science to the goal of mapping the human brain, specifically the gene activity in the brain. The scale of the goal is admittedly audacious - there are some 20,000 genes at work in the human brain, and plotting their activity in a 3D model will take 4 years and cost $55 million. But this is, at root, simply basic science, rather than cutting-edge research. It's somebody saying 'wait, let's take a step back and understand what this brain of ours actually does'. That sort of time-out for core principles is all too rare these days - everybody's eager to make a big score, the breakthrough, the home run. Nice that some people are still trying to lay out the baselines.

Is Our World Getting More Toxic?

An astute observation by the WSJ Health Blog that there's much ado about toxic substances these days prompted an informative Q/A with George Corcoran, president of the Society of Toxicology. He makes an interesting point - one reason we're seeing reports of toxic substances, such as pharmaceuticals in drinking water, is simply that we're now able to *measure* such things at such a small level. Which raises the question: Are we measuring something that is now there that wasn't before, or have there long been such background levels of toxic substances, but now we're simply seeing them?

A Whole Lot of Health 2.0 Companies

Some quick observations from the Health 2.0 conference in San Diego. 1) These panels are chock full of flashy startups. There's a profusion of companies at seemingly every step of the health care system. Reminds me very much of a a decade ago, the first dot-com boom. Each one seems to offer a promising tool, helping patients find a physician, say, or helping a patient narrow in on a more precise range of information for their illness. All useful tools.

But there's a great deal of overlap, and a great ambiguity of service (quick: what would a company called Carol promise to you? How about Xoova?). That's exciting, in that there's a profusion of services and innovation. But for a patient, it's likely confusing, and carries a risk: What to do if you spend months building a profile in a website, and then that company disappears (gets acquired, runs out of funding, changes business models, etc)? And for an investor? Eesh, what a crap shoot...

How We Know What We Know About Health In the U.S.

One of the great accomplishments of public health in the US is simply the existence of the thing in the first place. That is, Americans actually know something about our health on a population level. We take the availability of statistics and health data for granted; every day we read about the 80 million Americans with heart disease or the 8 million with diabetes and we don't really consider where those numbers come from. The significance of this information was highlighted for me last year, when I spoke with Christopher Murray, now the director of the University of Washington's Institute for Health Metrics and Evaluation. Funded by a $105 million grant from the Gates Foundation, the Institute follows a simple idea: If we want to improve health, we have to have a baseline understanding of what health is, and a scientific understanding of what works. Unfortunately, in most of the world, Murray notes, there's little knowledge about the most basic measures of health: mortality rates, birth rates, morbidity rates - these measures that we in the US take for granted remain question marks in many parts of the world.

So how is it that we know this stuff in the US? What mechanisms do we have for getting these numbers and validating their accuracy? Basically, the Centers for Disease Control and Prevention goes out and counts.

I was reminded of all this yesterday when I was walking along San Francisco's Embarcadero to visit the Ferry Building with my son. Right there on the Embaradero are four semi trailers, with the CDC logo and the words "National Health and Nutrition Examination Survey." Here's a picture (that's the Bay Bridge in the background):

Walking over, I thought this might be a kind of Bookmobile for health, where you can stop by and get some pamphlets or something. They weren't open, though, so I had to go online this morning to figure out what NHANES is all about. Turns out this is a much bigger operation than I'd thought. Turns out the survey has been going on since the early 1960s and has logged over 140,000 examinations. That's right, examinations. Here's a diagram of what's inside those semi-trailers:

So it's not a drop-in center at all; the CDC decides to visit an area, then plops its trailers in a central location and mails invitations to various people, based on demographic surveys. Though the CDC points out that this isn't a substitute for a regular doctor's exam - indeed, the tests aren't usually part of a standard physical - the information is useful for individuals, including results of STD tests, bone-density measures, kidney function tests, pregnancy tests, toxic exposures, and so on. Exactly which tests somebody will get depends on their age, sex, and so forth. Here's a video tour of the trailers: link

What comes out of NHANES are many of the metrics that pop up all the time: as the CDC says, "NHANES data have been used to influence policy and improve the health of the U.S. population in many ways including: getting lead removed from gasoline; creating and updating the pediatric growth charts; and establishing national baseline estimates for cholesterol, blood pressure, and Hepatitis C in the U.S." It all gets funneled over to the National Center of Health Statistics.

This is by no means sexy stuff; but it is a window into the workhorse engine of public health that is the CDC. Basically, what happens in these trailers becomes the baseline for our national health policy. It sets the yardstick for the state of our national health, and lets CDC researchers - and anyone else who wants a look at the data - evaluate how our health might be changing, for the good or the worse.

I don't mean to overdo it, but I find this pretty impressive. Even as so many people out there, including me, shake the trees for new tools that will let us measure our health in new ways - stuff like assessing our genetic risks to using proteomic technology for early detection of disease - it's worth remembering that it still takes a lot of work and money just to measure the simple stuff.

I'm gonna see if I can't get a tour from the CDC and follow up when the trailers open in a couple weeks.

Cold Medicine & Kids: Why We Should've Known Ages Ago

There's been much hubub in the news over the past few months about children and cold medicine, with the emerging reports and studies coalescing into the realization that the young children - ages 2 to 5 - and infants and toddlers, probably shouldn't be taking the stuff. (The FDA recommended such for infants this fall.) A study out yesterday in Pediatrics bolstered the thinking. Researchers looking at emergency room admissions found that 7,000 kids a year were going to emergency rooms because of cough medicine (mostly kids guzzling the stuff on their own). This is good science, and I'm glad to see that the system, on one level, is effectively monitoring over-the-counter medicines and raising alarms.

But I've got a contrarian take, too: Why do we need science - in the form of long-term, retrospective epidemiological studies - to suss out these public health concerns? Consider the Pediatrics study: It drew on two years of US emergency room visits from 2004 and 2005. And here it is January of 2008, and the study is finally completed and published. That seems alarming to me - we're going back four years to discover that there was and is a problem. But what's happened in those intervening years? All those missed cases.

I'm not suggesting the study is ill-conceived or designed; rather I'm asking why we aren't drawing better real-time information from emergency rooms and other aspects of our health system in order to spot and flag this instances as they're happening. In other words, why don't we have syndromic surveillance systems - which the federal government has spent billions on this decade - that can effectively flag these issues in real-time, so that we can take some sort of public health action? Then, sure, conduct a retrospective study and validate what the real-time information is telling us. But it seems to me that waiting for the mechanism of traditional science to make it's case is 1) overkill and 2) a band-aid, analog approach to public health when we should be pursuing instanteous, digital solutions.

After all, this isn't a bizarre, unexplained disease cluster or such that could have loads of causal factors. This is pretty straightforward - kids are being hurt by cough medicine, and going to emergency rooms because of it. Isn't that the low-hanging fruit a proper, robust surveillance system should be flagging as a matter of course? Alas, syndromic surveillance has been too much about spotting terrorist threats - which nearly never happen - or things like bird flu - which hasn't ever happened here - and not enough about flagging actual threats to public health - which are happening all the time. The data is out there, floating around. It's not hard to tap and sift.