PatientsLikeMe: A Study in Online Community Issues

The latest New York Times Magazine has an interesting story on PatientsLikeMe, a startup that's bringing the Web 2.0 user-generated-content approach to building communities of patients who suffer from a common disease. They upload truckloads of data about their treatments and symptoms (which the site compiles into outstanding visualizations) and also join discussion boards tailored to their specific illness.

In one respect, the site is an outstanding example of how to design a modern web-based social networking site. Look, for example, at the combination of information displayed on a typical profile page, which mixes space for free-form personal expression with informative standardized graphics based on disease-specific fields. Both this patient's personal statement and his symptoms chart are immediately legible to fellow ALS sufferers. These become a backdrop for conversations on the site's forums (closed to non-members). As the story vividly explains, PatientsLikeMe has been quite successful forming vibrant communities.

The site also, however, raises a set of thorny issues that are by now standard fare for user-driven online communities. The Times story focuses on the potential displacement of expertise. In much the same way that, say, The Motley Fool threatens the privileged status of professional investment advisors, PatientsLikeMe poses a threat to the privilege status of doctors. Patients armed with data from patients like them can pressure their doctors to change their treatment. The article tells the story of one MS sufferer who learned that, "Contrary to what his neurologist told him years ago, 10 milligrams [of baclofen] wasn't the maximum does. It fact, it was at the low end of the scale." Now he's taking 40 milligrams a day. Enough incidents like this one and the phrase "doctor's orders" will come to sound strange indeed.

Of course, doctors have some good reasons to worry, beyond just status preservation. User communities are notorious not just for displaying the wisdom of crowds but also the ignorance of crowds. The article mentions a group of 34 ALS sufferers who "solicit[ed] lithium prescriptions from their doctors and coalesc[ed] into an ad-hoc clinical trial." N=34, with no blinding and no control group is not an ideal experimental setup. The information it produces may have some limited value, but will the community have the statistical and methodological chops to assess that value? As PatientsLikeMe scales up, its users will move from the relatively easy field of data sharing to more difficult problems of data interpretation.

Another classic problem that user-generated-content sites face have to do with the linked problems of data quality and privacy. If PatientsLikeMe becomes a big success, will unethical drug company reps create fake patient profiles to report that the pill of the month provides dramatic results? There are plenty of feedback mechanisms to deal with sock puppets and the overeager-but-misinformed, but their proper use and tuning is a difficult art.

There are also remarkable privacy issues. The article focuses on HIPAA, which PatientsLikeMe neatly circumvents by being a service that individuals choose to use, rather than a health-care provider. (Interestingly, one doctor actually says that physicians -- and thus, not patients -- should be the sole stewards of patient data, neatly illustrating why some people will be cheering loudly if PatientsLikeMe dethrones doctors.) But that circumvention comes at a price. Insurance companies, you can bet, will keep a close eye on the site, looking to learn details about current and potential customers. I predict outraged accusations of egregious privacy violations, similar to the ones hurled by college students infuriated that potential employers have looked at their keg-stand photos on Facebook.

PatientsLikeMe, like it or not, will have to mediate many of these conflicts. Its user agreement (which purports to ban harvesting data from the site, along with registration by true outsiders to its communities) and its business model (to sell data to pharmaceutical companies) give a clue as to its attitude: we own the data, and will control access to keep that data valuable. But that's yet another classic Web 2.0 problem: why is it the site, rather than the users (collectively or individually) that owns the information they provide? Indeed, to the extent that the data consists merely of facts about medical conditions, are there even any meaningful intellectual property rights in it that could be owned?

Interesting stuff.