PatientsLikeMe for Partners Blog

Epidemiological Research

PatientsLikeMe conducts epidemiological research as disease communities reach a significant membership population. Based on our site activity over the years, we believe the critical inflection points for conducting epidemiological research on PatientsLikeMe are close to the following:

• Capturing 10% of Rare Disease Populations
• Capturing 5% of Medium Prevalence Disease Populations
• Capturing 1% of Prevalent Disease Populations

Known and Potential Biases of PatientsLikeMe Data

PatientsLikeMe is constantly working on mechanisms to ensure the validity and reliability of our data. While PatientsLikeMe represents what is happening in the real world, we understand the need to recognize and reduce any bias that our patient-reported data may have.  Known biases across the site include the following:

1. We collect real world data
2. All data are patient-reported
3. All participants are those willing to share health information on the Internet

We use a case-by-case basis. We have demographic comparisons of our largest communities to the published literature (available upon request). These comparisons are difficult because often the data doesn’t exist, or published reports don’t provide much information that is directly comparable to other populations. When usable data is available, our communities often, but not always, skew towards being Caucasian, female and educated. For example, a well-known multiple sclerosis (MS) database has 73% females[1]. PatientsLikeMe has 80%. However, the CDC estimates that 74% of those living with HIV in the United States are men[2], but on PatientsLikeMe, 77% are men.

An important consideration at the beginning of any PatientsLikeMe project is to determine the known demographic characteristics of the disease population external to PatientsLikeMe. We then determine the characteristics of the population within PatientsLikeMe and use that information to guide decisions about the study design (e.g., if it is necessary to oversample certain demographic groups or to screen for specific exclusion criteria).

Also, it is possible that PatientsLikeMe users, who elect to share health information on the Internet, are different from the population at large.  After all, the average PatientsLikeMe is very proactive.  An example is provided by Todd Small, a MS patient who became a member of PatientsLikeMe in June 2007. He learned from PatientsLikeMe that he was taking the wrong dose of a certain drug:

“For years, I had always taken just 10mg of Baclofen. I was told long ago by my neuro that ‘too much Baclofen can cause weak legs.’ Then I sign up here. Take a peek at what you guys are doing and find out I don’t take enough Baclofen to deal with my symptoms. Give the neuro a call, no problem, and [I am now] much, much better.“

Thus, quantifying the differences, if any, between PatientsLikeMe MS members like Todd Smalls and the MS population at large is an integral part of our research.

[2] http://www.cdc.gov/hiv/topics/surveillance/resources/factsheets/prevalence.htm, accessed June 30, 2011