The Wikipedia Approach to Disease Classification

The International Classification of Diseases is one of those things that everybody needs, but nobody knows exists. In a nutshell, it is the official list of diseases, as determined by a collective body of experts, under the authority of the World Health Organization. It dates back to the 1850s, when it was known as the International List of Causes of Death, while today it is a highly codified (and oftentimes politicized) list of all sorts of ailments, from broken bones to a dozen or more forms of colitis (here's a PDF link to the latest updates, from 2005). Now on the 10th edition - ICD-10 - each new edition is bigger and better and (again) bigger than the last.

Each disease has its own specific code, and that code is gold – it's used by physicians to specify a diagnosis, by insurance companies to log payment, and by lobbyists to maneuver for new disease classifications (I wrote about one such effort in Wired last fall regarding metabolic syndrome - code 277.7 - and the stakes involved in recognizing that dignosis as an official disease). If a new condition is given a ICD code, it is well on its way towards official recognition as a disease, with all the trappings of insurance payments, industry conferences, and pharmaceuticals that that entails these days.

So I found it especially intriguing to read that the WHO is considering opening its selection process for ICD-11 into a public wiki. That means everyday folks can suggest new diseases and new classifications (this happened a couple weeks ago, but for some reason only the Canadian press covered it, at least as far as I could tell). Unlike Wikipedia, they will have top editors vett the new suggestions, so I think it may be partly motivated by PR considerations, or as an experiment that - I hope - is easily undone if things go awry (ie, if spoofers decide to elevate hangnails into a form of cancer, or some sort of mischief).

But it's an interesting reflection on the notion of disease, really. The implicit message here is that there are all sorts of conditions that deserve official recognition that the medical establishment is somehow missing or is ignorant of, and that by canvassing the people they'll turn up some ailments that deserve full ICD inclusion. Is that possibly true - are there really diseases (or pre-diseases) that the medical world isn't already tracking? And is this just a recipe for more "disease mongering", ie medicalizing the everyday annoyances of life? We'll see. But it should make for a more exciting than usual classification meeting this fall.

Biosensors: What's Out There, Really?

There was a great hubub back in 2003 after Pres. Bush announced the creation of BioWatch, part of the Federal response to the anthrax attacks of Oct. 2001 and the threat of bioterrorism. The objective of BioWatch is to deploy thousands of sensors across the country - mostly in cities - sensors that can pick up biological, chemical, or nuclear traces. Basically, the sensors are part of an early-warning system primed to detect pathogens like anthrax, smallpox, tularemia and influenza. They are assumedly agnostic as to the origin of a trigger - they can't detect intent or whether the release is from a bioterrorist or a Indonesian chicken that snuck over to Queens. (BioWatch is not to be confused with BioSense, a syndromic surveillance network that gathers intake information from hospitals, over-the-counter sales from pharmacies, lab tests, and other info into an ad-hoc information network. Nor to be confused with BioShield, the ambious effort to provide an end-of-market demand for the vaccines and pharmaceuticals that might be used to treat a biological attack.)

Information about the program is very, very sketchy (not even the RAND Institute could get clear figures for a recent white paper on Infectious Disease and National Security [PDF LINK]). Some estimates have said there are 4,000 sensors installed in 31 cities extremely quickly (in a matter of months) in 2003 - just where has never been made public. There's been some criticism about how the program has been run, first in 2005 from the EPA inspector general's office (PDF LINK), and then this year a little-noticed report from the Homeland Security IG.

And there have been some false alarms - most prominently a tularemia attack in Houston that turned out to be simply environmetal bacteria.

And it's been hard to even discover what the sensors are or how they work - this Washington Post story implies that they are the size of a refrigerator and the samples are collected then tested manually back at a lab (I assumed they're using some sort of DNA-based molecular diagnostics at the lab).

That makes the "sensor network" far less high-tech than it seems to be, and indeed helps explain why it has been so cheap - this 2005 story from Government Executive puts the price at $1 million for installation and $1 million/city/year for operating costs, putting it at about $200 million for the three years from 2003 through 2005. That's exceptionally cheap for a Homeland Security program, but also underscores the "on the cheap" nature of what's been done so far. In other words, this may be more band-aid than panacea.

It also gives some context to this press release today from Temple University, that heralds a biosensor researchers at Temple's medical school have created. Who knows if Temple's will make it to market (it sounds very experimental, judging by the release). Still, thank god they and others are still chasing after improving these things. We may already have a network deployed. But that doesn't mean it's actually working.

Stepping Beyond Personalized Medicine

One of the big consequences of the germ theory and the move towards a more scientific medicine was the consequent move away from the way medicine had been practiced. That sounds obvious, but here's what I mean: Scientific medicine (which is pretty much how we're all treated now) was the systematic and systemic use of rigor and experiment in medicine, so that some universal facts could emerge and patients could be treated for widespread, rather than individual, conditions. Meaning disease could be treated broadly, for many people at once, rather than piecemeal, each person to their own symptoms. This was distilled into the aphorism: Treat the disease, not the patient. The old way of treatment was called - and I love this term - idiosyncrasy (note the wikipedia link!), and it held that every patient's illness was idiosyncratic, and thus must be treated distinctly. Needless to say, this was very inefficient. But with a greater understanding of the broader causes of disease (especially infectious disease, via pathogens) and a greater opportunity to see disease before it manifests as symptoms (via X-rays or microscope), scientific medicine let the medical industry treat people in far greater numbers, with far better results. Which brings us the age of medicine and public health we have today.

Except: What's intriguing to me these days is how we're moving back to an age of idiosyncrasy.

This is, of course, usually called personalized medicine (Wikipedia link!), and its generally understood as the ability to use a patient's genome to understand and treat conditions they may have unto themselves, via pharmacology (pharmacogenomics is the term of art here). And as much as "personalized medicine" sounds like empty consumerist jargon, I think it will establish itself as a primary mode of care (more expert people than myself think so, too).

But here's what's really cool. No sooner will this idiosyncratic model remerge, than it will be superceded by a kind of collective idiosyncrasy: the use of large genome databases to extrapolate the notions behind personalized medicine onto a population scale. I'm thinking here of the HapMap project and many others.

I'm going to post more about these projects in coming days, but in the meantime - while I may have a few stray Slashdotters poking around - I'm eager to gather thread on the many diverse human genome databases and related consortia out there, I'm thinking of everything from the original National Human Genome Research Institute to things like the Cancer Biomarker Consortium at the Hutchinson Institute I mentioned the other day. If you have a favorite, drop it into the comments. I'll be reemerging with a bigger thought on them soon.

Welcome, Slashdot Readers!

So I got slashdotted. First, welcome to all of you, and thanks for reading. I knew opining on Wikipedia would create some attention, but I didn't quite gauge the scale of things. Anyway, some quick reactions:

1) I do know the difference between 'mitochondria' and 'mitochondrial DNA' - the NIH quote from that post is from a page entitled "mitochondrial DNA" and the point is to look at the results from a Google search of "mitochonrial DNA" - Wikipedia's first, and NIH is fifth.

2) When you critique Wikipedia, a few things are gonna happen:

a) ...the Wikipedians will come out and say: "Don't complain - edit!" Sorry, that's not my job. I do edit when it's something I know about (I've lent a hand to entries on the Replacements, Queen Elizabeth, Petrarch, and metabolic syndrome, among others). But when I look to Wikipedia to learn about something - ie, when I use it as a reference, not as a 'project' - I use it to understand a topic, not to help create the resource. You wouldn't want the ignorant likes of me editing those entries, anyway, right?

b) ...mention a few pages that are written poorly, and - brilliantly - they will be improved summarily. The epigenetics page has been nicely cleaned up in the past 24 hours. It doesn't debunk my argument, but it definitely wouldn't be exhibit A for me any more - and yes, it does show the power of Wikipedia.

c) ...people will accuse you of being ignorant, stupid, and a dumbshit. Whatever.

d) ...people will say you're trying to dumb down Wikipedia. Look, I make my life as a journalist. What we do at Wired is translate difficult topics - topics of urgency and significance but not necessarily ones that our readers would come across otherwise - so that people can understand what they are and why they matter. Whether you individually think we succeed or not, I can't control - that's for our readers to decide collectively. So far, they decide we do it pretty well. We've got 650,000 readers, two National Magazine Awards in past three years, etc etc. We are useful to many people. Same goes for Wikipedia. It is, like it or not, a filter, a tool that's for the benefit of readers who want to know about the topics it includes. I am such a reader. I have observed that, often, it defaults to an academic and impenetrable style when, in fact, some clarity and explanation would serve those readers well.

You may be a microbiologist, or a mathemetician, or a geneticist, and you have no problem understanding these entries. Congratulations - you are officially smarter than me and many, many, many other users of Wikipedia. But Wikipedia's not your little playground. It exists as a resource used by everyone; that is its power. Am I saying “dumb it down!”? No. Am I saying science is hard? Yes. But it’s not impossible to write clearly, nor to help the curious become the informed. With the collective resources Wikipedia has - even, at times, including myself - I think that it would be nifty if the writing could be clearer and better.

3) There are thousands (!) of new readers to this site now. Some of you will never return. But I hope some of you do, because I think what I'm trying to do here is what I'm asking of Wikipedia - trying to make sense of some very important trends in science and health and medicine, and trying to help a broader base of people understand them. It's great that there are scientists out there working on important things - genetics, biology, on and on. But I believe the true, full potential of these disciplines and their research lies in helping the broadest number of people understand it and appreciate it (indeed, if it's research funded in part by the government, it's incumbent on those scientists to help those citizen-patrons understand it).

So check back in. You may not agree with everything here. And it's not going to be definitive. But at least it may be from time to time- as it already seems to be today - a bit provocative.

Why Does Wikipedia Suck on Science?

Wikipedia is, by all measures, one of the great accomplishments of the Internet Age. I'm willing to say it stands alongside Google, eBay, GoogleMaps, IMDB and Wired.com as among the greatest resources on the Web (ok, that last one is self-serving).

But boy, does it suck when it comes to science topics.

Here's my beef: Wikipedians are at their best when they are able to use their knowledge, be it bonafide expertise or particulates of trivia, to fill in the blanks for our collective intelligence. There's nothing like needing to know who Henry V was (versus Henry IV) and being able to find the answer in less than 50 keystrokes. Or pinpointing just when the first Anthrax attack happened in 2001. Or getting a good sense of whether Francis Fukayama is a neocon or a brilliant maverick (mutually exclusive??).

But I find that when it comes to science topics, I often find Wikipedia more of a hinderance than a help. Curious about just what epigenetics is? Figure you really should know what mitochondria do? Don't count on Wikipedia - odds are their analysis is too pedantic for you, as it is for me.

Now I'm no Wikipedian come lately. I wrote the first story in Wired Magazine on Wikipedia about four years ago, back when it had a paltry 150,000 entries in English (it boasts 1.8 million and climbing now). But it's an interesting problem that seems to arise when you task experts to write on an expert topic. When you're open sourcing Linux with programmers, the fact that they're all speaking the same language - or writing in the same code - is a benefit. But when you're creating a quasi-open-source project for, well, everyone, it may just happen that the expert langauge necessary to define a topic will progressively escape the comprehension of the non-experts who are the main audience for said project. It's not quite forking, to use the open-source term for when a project gets split and "forks" into tangential projects. It's more like oyster forking - the creation of a highly specialized tool that only some people can grasp.

Here's what I think is going on: On Wikipedia, contributors are expected to contribute their knowledge. But on science, there's a oneupmanship going on, and a topic will be honed to an ever-greater level of expertise. That's great for precision and depth, but horrible for the general user, who is often brought to Wikipedia through a top hit on Google. Clay Shirky and others have written about the "the expert problem" on Wikipedia, usually meaning the lack of expertise and a need for experts. That may be true in some contexts, but that isn't the problem I'm talking about. That complaint is that Wikipedia needs experts to bring entries up to snuff; I'm more concerned about bringing entries down to a level that's actually clear and useful for the layman.

Look at that Epigenetics entry, for instance, which comes up first when you Google the term "epigenetics". Here's the first sentence:

In biology, while the subject of genetics focuses on how organisms can inherit traits by inheriting genes from their parent(s), which encode information for cell function as sequences of DNA, epigenetics is sometimes used to refer to additional methods of biological inheritance that do not directly relate to the inheritance of collections of genes, or soft inheritance.

Huh?

Now I'm sure that's accurate, but it's way too rich for my blood. A better primer can be found at the backgrounder from Johns Hopkins that ranks as the number three hit:

There is far more to genetics than the sequence of building blocks in the DNA molecules that make up our genes and chromosomes. The "more" is known as epigenetics. What is epigenetics? Epigenetics, literally "on" genes, refers to all modifications to genes other than changes in the DNA sequence itself. Epigenetic modifications include addition of molecules, like methyl groups, to the DNA backbone.

That, I get. It's the same on so many other topics. Here's the first line for the entry on fluid mechanics:

Fluid mechanics is the subdiscipline of continuum mechanics that studies fluids, that is, liquids and gases. It can be further subdivided into fluid statics, the study of fluids at rest, and fluid dynamics, the study of fluids in motion. Modern applications use the computational approach to develop solutions to fluid mechanics problems; the discipline concerned with this is the CFD, Computational Fluid Dynamics.

Sorry, you lost me at "continuum."

And here's the beginning of the Mitochondrial DNA entry:

Mitochondrial DNA (mtDNA) is DNA that is located in mitochondria. This is in contrast to most DNA of eukaryotic organisms, which is found in the nucleus. Nuclear and mtDNA are thought to be of separate evolutionary origin, with the mtDNA being derived from bacteria that were engulfed by early precursors of eukaryotic cells.

Thank god for the NIH, which helpfully has a page (the fifth result from Google when you search for "mitochondiral DNA") that starts with this:

Mitochondria are structures within cells that convert the energy from food into a form that cells can use. Although most DNA is packaged in chromosomes within the nucleus, mitochondria also have a small amount of their own DNA. This genetic material is known as mitochondrial DNA or mtDNA.

Thank you, big sprawling federal bureaucracy!

Now given that there are, as I said, 1.8 million articles on Wikipedia, there are bound to be dozens - if not thousands - of exceptions. For instance, on the basic science entries - biology, cancer, volcanology - Wikipedians have created useful, thoughtful, and readable dispatches. And sometimes there's been the laudable foresight to add "introduction to" pages, such as those for quantum mechanics and quantum physics. But increasingly, I find myself skipping over a Wikipedia result on Google not because I'm worried about the validity of the information there - I don't share that concern and think it's way overblown - but rather because I'm worried it's just going to be a bunch of formulae I can't parse and jargon I can't unpack.

This is, in many ways, the opposite of the tragedy of the commons - it's the tragedy of the uncommon, meaning topics that the common folk just don't get - and thus can't help in editing the entry on. What happens when you get something written by a bunch of geniuses? Well, something written by a bunch of geniuses.

This Week in the MMWR: Education & Health

One of the most striking correlations in all of public health is the strong association between education and health status. The more education people have, the better health they seem to have. For a striking graphical look at the phenomenon, the CDC's MMWR this week has a great chart on self-reported health status and educational status. "Association does not mean causation" is a mantra of epidemiology, though - meaning that just because there's a link that doesn't mean more education causes improved health. It could be any number of confounders. Basically it's a chicken/egg thing: Are the better educated likely to have better health because they're in school - or are the more healthy simply more likely to get an education? For health economists, this is an endless source of chatter and speculation. They can look at all sorts of side effects: Are the better educated more likely to learn about sound health practices (ie, don't smoke)? Are the better educated more likely to earn more, and therefore be able to buy better health?

That subtlety between cause and effect is usually missed, though; the New York Times did a big front-page story in January headlined "A Surprising Secret to a Long Life: Stay in School." (The story is stuck in the paid-only archives now, so I'll omit a link).

Ultimately, of course, it may not matter what's behind the connection. If going to school teaches you to live healthy, great. If healthy people are more likely to stay in school, great. Either way, you'll end up healthy and wise - if not necessarily wealthy, too.

A Test for Every Genetic Disease

One of the challenges of tracking the future of healthcare is the signal-to-noise problem: seemingly every day there's a new report from a lab that claims to have isolated a gene X for condition Y. These reports (usually press releases from a university PR dept) carry the requisite "so what?" paragraph that goes something like: "It is expected that in the future, this work can be extrapolated into clinical tests for condition Y, helping physicians diagnose earlier and helping patients get treatment sooner." Keith Robison's Omics! Omics! blog has a much greater tolerance and acumen for this stuff than I'll ever have, so you won't see me try to keep pace with it all. But read them for months on end, and these "in the future..." paragraphs start to seem like a shellac layered on by said university PR department, in order to give their research a relevance that they really don't yet have. After all, these genes are being isolated in labs, in vitro, and are usually one-offs - a long way from replicable results.

At least that's how I read them, skeptic that I am. Until I happened a couple weeks ago across GeneTests.org, an NIH-funded directory of genetic tests that are available now, today. If you thought that genetic testing was only a piecemeal phenomenon so far, and largely experimental at that, GeneTests is an eye-opener. There are, in fact, over 1,300 tests now available for various diseases. Most of these are for detecting inherited diseases, ones that are largely binary in nature - you have the gene, you have a high probability of developing the disease. These are Huntington's, Alzheimer's, and other diseases that have very clear genetic origin.

Hsien Lee reminded me of the site in a recent post endorsing the issue backgrounders.

For a non-geneticist, I think it's striking to recognize how significant it is that we already so many hundreds of genetic tests already available. What's more, this is only the low-hanging fruit of genetic testing - the still more revolutionary stuff is when we can start crunching the presence of genes, the first signs of biomarkers, and certain environmental/behavioral conditions that get at more elusive diseases. This is the stuff of genetic epidemiology, such as the International Cancer Biomarker Consortium that Lee Hartwell is cooking up at the Hutchinson Cancer Institute. And my hunch is that this stuff is gonna be huge.

This Week in the MMWR: The Smallpox Vaccine Spreads a Pox

A side-effect of vaccines has always been that, in some people, the treatment actually induces illness rather than immunity. After all, the idea behind vaccines, ever since Edward Jenner, is that exposure to a teeny bit of a pathogen is enough to kick in the body's antibodies and develop resistance, but not enough to actually foster disease. There's always been some imprecision in this, though, given that vaccines are deployed across huge populations - some number of which will always have lowered immunity or other risks that will mean even a small exposure is enough to germinate disease. That's all backstory to this week's MMWR report from the CDC, which includes a case of a woman who developed a strange infection after having sex with a Army soldier who'd recently had a smallpox vaccine. The genital infection wasn't herpes or any other common sexually transmitted disease. This was a second-hand infection, meaning she developed a disease even though she hadn't herself gotten vaccinated (thus making for a particularly tricky outbreak investigation).

At first glance, this is terrifying: If the woman got infected from a smallpox vaccine, does that mean she has smallpox? And since that diseaes was famously eradicated, are we at risk of spreading than bane of mankind all over again?

Thankfully, no on both counts. The virus used in smallpox vaccines isn't actually attenuated smallpox - it's closer to a strain of cowpox - vaccinia - which is similar enough to smallpox to foster immunity, but not nearly so lethal as the smallpox virus proper. Interestingly, this is the second case of secondary infections from smallpox vaccines that I've seen in recent months. In March, a child developed a dire case of eczema vaccinatum, a rare type of vaccinia virus skin infection, after contact with his father, who had recently been vaccinated against smallpox.

Oh, and you may be wondering: If smallpox has been eradicated from the face of the earth, why are we still vaccinating military personnel for the virus? Good question. It seems that even though the virus officially exists only in two, highly guarded places - at the CDC in Atlanta and at the Russian Vector State Research Center of Virology and Biotechnology in Siberia – there's reason to believe the Soviets weren't content to just let it sit under lock and key, and instead developed smallpox as a possible bioweapon (there's a terrifying book by Ken Alibek that describes the Soviet effort to weaponize smallpox and other deadly pathogens - Biohazard).

Chance enough that as of 2002, all personnel are now vaccinated. So while there's no sign of smallpox out there (so far), there will continue to be these little eruptions of vaccinia, because of the vaccinations.

Big Big Study

The American Cancer Society recently announced it needed some volunteers for an upcoming prospective study on cancer. How many volunteers does it need? 500,000.

That's a big study. The idea is to find Americans between 30 and 65, with no cancer, and track them for the next 20 years through blood tests and questionnaires. This Reuters story says it matches some similarly sized trials in Europe and Asia. But it makes me wonder: What's the largest cohort ever assembled for a forward looking study? (It's easier to do retrospective studies with big numbers because there's no tracking to do; it's all done through records). Here's a prospective study in India that claims 14 million participants, but it looks like it's all done through vital statistics records. In the UK, Biobank and the EPIC study both claim half a million participants and make claims of "most participants" status.

It reminds me of my favorite clinical study, the legendary Framingham Heart Study, where the National Heart Institute recruited just about the entire town of Framingham, Mass., to participate in a study on heart disease. The study started in 1948 and continues to this day; it's now tracking a third generation of citizens (and is on at least its third generation of researchers). Unprecendented for it's day, Framingham tracked just over 5,000.

GINA: The Coming Law on Your Genetic Secrets

Lots of excitement this week on GINA, the Genetic Information Non-discrimination Act. On the whole, I think this is a good step, and a good thing to get out ahead of what I believe will be a wholesale shift in our approach to health. But I don't want to repeat what's been said elsewhere, so I'll link to the good and thorough GINA Primer over at The Genetic Geneologist.