The Other Upside to Electronic Health Records

For those readers who may've seen my op-ed in Wednesday's NYTimes on electronic health records, there's one upside that I didn't mention.

The predominant argument for EHRs - including mine - is that they'll make for better individual health care and, system-wide, can save lives and money. What's more, if a national standard for EHRs emerges, such as WorldVista, the system I wrote about, then the US health care system could in fact be faster, more efficient, and more connected as well.

All that is true, and it is, to my mind, reason enough to spur the government (via CMS or some other HHS agency or program) to promote the adoption of EHRs as soon as possible – like now.

But there's one more great side effect of EHRs - the data.

If the US health care system was to move to nationwide electronic records, and if those records were largely standardized and exportable, there would be an unprecedented reservoir of health data for epidemiologists and researchers. The benefits wouldn't just be statistical or demographic - who's dying when of what, etc. Think bigger: With information on millions of Americans health histories, we could ascertain what sort of treatments work - and which don't. We could see what the optimal time for interventions is, and when is too late or too early. We could see where in the country we are doing things right, and where things are going wrong (who knows - maybe Mississippi has some health secrets that aren't turning up in the rudimentary record-keeping we now have).

If you're wondering about privacy concerns, I have two answers. First, the cheeky one: Get over it. I think our national obsession with privacy, especially on medical histories, is overblown and self-defeating. Other countries don't worry nearly as much about privacy issues, they have set up remarkably rich electronic health systems, and they're already pulling out reams of great data that help them treat their citizens better. (Want to know whether autism is related to vaccinations? It's not - Sweden proved that already, thanks to their swell record-keeping.) And that's not even counting the upside once genetic information starts dropping into the system.

Second, the pragmatic answer: Stop your worrying. We have plenty of laws that provide for privacy of health records (HIPAA, etc), and if anything they're too strict about the stuff. What's more, the data I'm talking about can be completely anonymized. We don't care what's wrong with Joe Brown from the Upper West Side – we care about what's wrong with the Upper West Side.

Right now, all that data is just slipping away, and researchers have to look to other countries - countries that are smaller, less diverse, and differently structured - and awkwardly extrapolate to the US. I talk to researchers about this - especially data wonks - and at first they get so excited about the potential, about all the studies they could do, all the questions they could answer. But then they wake up - they're resigned to getting those answers elsewhere, or forgoing them, because we simply don't have a mechanism to turn what your doctor does for you into data that can be combined with what everyone else's doctor is doing for them.

The fact is, the US has the best medical care in the world despite itself. We don't really know what we're doing, or what works, in any really scientific way. Sure, we have studies that show - under very controlled circumstances - what sort of treatments work. These demonstate statistically significant results under laboratory conditions. But in practice? In the real world of clinical practice (small scale)? In the real world of health care practice (large scale)? It's a total crapshoot. In the worst case, that's partly why we have these massive after-the-fact pharmaceutical recalls; because we (they) didn't really know what was going to happen once the real world scenario started to play out. The data simply isn't available.

This is the real bonanza of electronic health records in the US. The data become available. But let's be honest: The fact is, it's an argument that won't change anything, won't ever get anything done, because there's no personal benefit. It's all in the abstract. But gee, it'd be great to get there...

And while I have you: Like what you're reading? Sign up for the RSS feed.

The New York Times Op-Ed

Today (Wednesday) the New York Times published my op-ed on electronic health records, specifically on the open-source records system called WorldVista. My take: it could be the breakthrough backend platform that the US health care system needs to FINALLY go digital. Here's the link, and here's the gist of the story:

Health care providers have been dreaming about electronic records for so long that the idea has begun to seem like vaporware, a never-to-be-realized fantasy similar to flying cars and jetpacks. But there is already a clear software standard, an open-source system that’s low-cost, easy to use and readily available. It could be the key to the health care system we ought to have already....For the vast majority of health care providers, WorldVistA is what they’ve been waiting for: a low-cost, simple-to-use system that makes it easier to provide quality health care.

Want more details? Please see the subesquent post for more.

The Return of Quarantine

This story on the wires today: For the first time in four decades, the CDC has quarantined a man infected with extreme-drug-resistant tuberculosis. The item provokes three thoughts: 1) It's a growing truism that the ease of international travel is a boon to infectious disease (see SARS). But that doesn't make the ease of aviation-propelled infection any less alarming. And the real appeal isn't just that airplanes make it easy for a pathogen to skip across the globe - it's also the luxurious incubation time that close-contact airplane travel provides (what virus wouldn't love 200+ humans sitting inches from each other and swapping breath for 9 hours?), so that when the original carrier lands, his fellow passengers will help him finish the job.

2) Quarantines are an ethical issue waiting for a 21st century reconsideration. In the last few months, there have in fact been several stories like this one - Americans who turn up with TB and are quarantined, sometimes against their will - sometimes with the possibility that they'll die there. The difference now is that those have been almost invariably handled as a local matter, while in this case the CDC is at work. So to me, that means the issue of quarantines - when to do it? to whom? under what circumstances? - beg for some sort of clarity.

So what's the prevailing policy out there? A quick search of the CDC website reveals a rather officious, distant tone on the subject (much of it seems related to SARS, which happened four years ago). At the other extreme, some civil libertarians are all too ready to scream outrage at the police state - Is Sickness a Crime? they shout. Well, as the 100-year-old case of Typhoid Mary reminds us - yes, sometimes.

There's been some new thinking on the ethics of quarantine policies in policy and academic circles, specifically in relation to pandemic influenza and XDR tuberculosis, but ethics is different than policy. By and large the rhetoric on the subject seems outdated and out of step with the prospect. And if a pandemic does happen, and given that we are rather on edge these days, this could be a policy crisis in the offing. 3) good thing he was headed to Georgia anyway, where the CDC happens to be based (bonus points for those who know why - no fair Googling the answer!).

The End of Cancer

"It's not going to be easy." So says Dr. Lee Hartwell, the director of the Fred Hutchinson Cancer Center and chairman of the scientific advisory board at the Canary Foundation, re: the goal of ending cancer. But that doesn't mean he doesn't have a strategy for getting there. I'm at Stanford, sitting in on the annual Canary Found. Symposium, a gathering of about 100 stellar  researchers and scientists who are all focused on the early detection of disease (thus the name "canary"). The guiding principle is simple: If we can detect disease early (specifically cancer, from the Canary's POV), we can save thousands of lives. Canary founder Don Listwin is a maverick Silicon Valley veteran who offers the goal in one graphic that looks something like this:

That's my rough redo on his slide based on a quick glance, so don't those numbers as anything accurate. But the take-away is obvious: Intervene early, and you have a good chance of survival. Intervene late, and survival rates are slim to none.

The "not easy" things Dr. Hartwell is talking about start with proteomics, the search for protein biomarkers that signal early (or even pre-stage) disease, but they quickly multiply. The challeneges are vast: What are the proteins we need to look for? How do we find these proteins (and more specifically, how do we find them cheaply, easliy, and without an invasive procedure)? Having detected the biomarker, how do you locate the cancer in the body? And having found it, how do you treat it? Do you surgically remove it? Or do you simply leave it alone (if it's a non-fatal cancer)?

Each of these questions requires a vast deployment of science, innovation, and luck to answer. And each answer must not only be answered, but systematized, to create a viable, FDA-approved aresenal of diagnostics and treatments. As Anna Barker from the National Cancer Institute puts it, this is most definitely "big science" - the sort of goal that takes great coordination, great patience, and great amounts of money. But the upside is huge. She describes a medical model where we get our genome sequenced at birth, along with an assortment of other baseline marking. Then as we grow we follow specific behaviors and practice preventive techniques suitable to our profile. And as we move into adulthood and age, we undergo routine screening for those conditions we might be susceptible to, and if necessary undergo early treatments. At each stage, we're improving our odds for longer life.

It sounds easy. But it's not going to be.

More later.

Economies of Scale, Meet Sequencing

There's lots of attention brewing up around the X Prize for Genomics, with the target of getting the price for a human gene sequence to $1000. At that price point, a whole lot of things become possible that aren't now. Like, say, getting your DNA sequenced with your annual checkup, etc. The X Prize model is terrific - Wired gave Peter Diamandis a Rave award in 2006 for taking it from space to sequencing. But it's sometimes seems too futuristic; of course sequences will cost $1000 some day, but that reality is very different from today's. And until then, what?

Well, look a little deeper and it turns out that day isn't quite so distant. Consider this neat post from Derek Lowe's In the Pipeline, on how some researchers were treating a patient with a drug-resistant staph infection by sequencing the bacterium's entire genome to watch the mutations. Lowe provides some nice perspective on how holy-cow this really is:

Here's natural selection, operating in real time, under the strongest magnifying glass available. ... A few years ago, needless to say, it would have been a borderline-insane idea, and a few years before that it would have been flatly impossible. A few years from now it'll be routine, and a few years after that it probably won't be done at all, having been superseded by something more elegant that no one's come up with yet. But for now, we're entering the age where wildly sequence-intensive experiments, many of which no one even bothered to think about before, will start to run.

A $200 Million Movie Idea

I have an idea (based on genetics) that is a slam-dunk plot line for a Hollywood movie. Far as I know, it is a totally original, and totally plausible idea, worthy of a thriller or a quasi-futuristic action movie (the sort that make $200 million or more). It will make for both a good plot and a dramatic and unanticipated twist ending. I am willing to discuss it with serious prospects. (Thanks to Chuck for the template here.)

Why Genetic Medicine Is Tricky

So there's a great deal of enthusiasm here and elsewhere - such as this swell DNA Network I'm a part of - for the power of genetic based medicine. The overarching idea is temptingly simple: Once we can spot the genes that cause disease (whether individually or in combination) we'll be able to intervene earlier and treat better. On paper, it's a pretty plain proposition. All we have to do is map out the range of genetic traits and preconditions, and we'll be able to maximinze the good genes and minimize the bad ones (this is, of course, a gross simplification). But:

Turns out this is harder in practice than in principle. Consider this recent news about a discovery at the Georgia Institute of Technology that turned upside-down a commonly held perception of one particular gene. Published in Plos One, the gist is that a gene scientists thought helped chemotherapy destroy cancer cells actually may, in fact, help cancer thrive. Here's the lowdown:

When a cell is malfunctioning or injured, the gene p53 is called into action and tries to repair the cell. If the cell can't be repaired, p53 starts a process known as apoptosis that kills the cell. It's p53's role as one of the genes involved in initiating cell death that has led cancer researchers to long believe that the gene is essential to successful chemotherapy. The idea is that p53 assists in killing the cancerous cells that the chemo treatment injures.

But in this latest trial, Georgia Tech researchers found that p53 may be a "double-edged sword." Chemotherapy patients whose tumors had a mutated p53 gene that didn't work had a much better survival rate than those who had normal p53.

So here's my take-away: We may be moving forward fast on mapping out the varieties of the human genome. And we may be quickly learning which genes do what and why. But genes are very tricky things. They don't act, as they're often described, like computer code - that is, in a binary either/or way. Rather they have multiple characteristics with multiple effects. The idea that we can simply turn certain genes on - to improve good effects - and turn other genes off - to reduce ill - is a superficial, and ultimately naive approach. In other words, it's going to take much longer to work all these characteristics out than is commonly described.

AJ Soprano's Food Conspiracy Theory

A great moment in tonight's Sopranos episode: when AJ, before worse things happen, rants at a family dinner about how meat is unsafe because the FDA is letting companies spray viruses onto the meat to eliminate bacteria. I don't know about you, but I let out a huge: "Huh?" Turns out AJ was right: last August, the FDA approved a cocktail of bacteriophages that attack the bacteria that causes listeria, a classic food-borne bacterial disease that can sometimes be fatal (500 people a year in the US die from listeria). The spray is developed by the biotech company Intralytix, and according to their website its a totally safe additive that can only add to a reduction in food-borne illnesses.

The sceptics say that there's too much unknown about the properties of the phages, and that - on the face of it - putting viruses onto foods is a bad idea.

So: is it a conspiracy theory? There's sure a lot of crazy websites out there: here's a Google search - most of the hits are to fringe nutritional websites. And here's the FDA's FAQ about the approval.

So if nothing else, AJ was definitely on to a genuinely deep conspiracy theory. Kudos to Sopranos creator David Chase for catching this one.

Pathogens, Pathogens, and More Pathogens!

In more database news, a neat new project out of the University of Maryland was announced in Plos Computational Biology today. Called Insignia, it is, in their own words:

a project to create and store information about a broad range of microbial pathogens, focusing primarily on bacteria and viruses. There are two major components to the system:

-A computational pipeline to generate unique DNA signatures for any and all pathogens in our database -An integrated database containing genome sequences, phenotypic information, and comparative genomic analysis of all pathogens

They've got over over 3,000 organisms in the database (though it's not clear whether there are full sequences for all of these). I'm not sure on the face of it how this is different than the National Microbial Pathogen Database Resource, which NIAID set up in 2004. Both seem to be positioned as potentially one-stop-shops for the new generation of diagnostics, tools that will be able to quickly scan samples looking for snippets of a sequence for a rapid diagnosis.

I've pinged the lead researcher with a couple questions, and will follow up when I hear from him.

UPDATE: So here's what Adam Phillippy from the Univ. of Maryland's Center for Bioinfomatics & Computational Biology (rolls off the tongue...) adds:

Hi Thomas,

There are many great resources out there for hosting pathogen metadata, but Insignia is the only one I am aware of dedicated to signature design. NMPDR offers a “signature genes” service that appears similar to Insignia, but only identifies unique genes and is not as well suited for assay design. For instance, Insignia can design probes for differentiating between two nearly identical strains of bacteria, whereas the NMPDR system would be unable to do so. Like you mentioned in your blog, we aim to be a one-stop-shop for diagnostic assay design. An assay designer could draw sequences from our database and use them directly in their diagnostic. That is exactly how we designed our Vibrio cholerae assays, and they proved to be quite accurate.

Insignia is definitely a work in progress. We are only a year into what is planned to be a three year development effort. We do encourage others to contribute sequences to our database, and to the entire project as well. We freely release all of our code and experiments.

Thanks Adam.

As I said, very cool stuff.

Aggregating Genetic Knowledge

Cool new resource/clearing house for genetic databases and projects: the Genetic Alliance. It's a database drawing together some 600 organizations regarding over 1000 genetic conditions. Many of these, no doubt, are conditions few people have ever heard of: Trichorhinophalangeal syndrome? Refsum disease? The goal is to draw on the collective power and knowledge of the organizations that spring up for these relatively obscure disorders, so that they may more effectively lobby, educate, and spur research. Interestingly, it's funded in part by the CDC.