This article is a bird’s-eye view advocacy piece for anyone interested in the Inflammatory Bowel Disease community.
There’s a difference between the drugs commonly prescribed for IBD (Crohn’s and Colitis) and the set of drugs commonly taken by IBD patients. A simple example is to consider an IBD patient suffering from abdominal pain and can’t sleep. A typical course of drugs would not only be pain medication but also sleep medication. A few resources on the Internet do an adequate job of discussing the drugs taken for IBD symptoms and remission maintenance but there aren’t any resources which examine the distribution of drugs taken by IBD patients. This article attempts to quantitatively examine the distribution of drugs taken by IBD patients. I’ve done a meta-analysis using what patients contribute to a popular IBD discussion board called crohnsforum.com. The aim is to count the mentions of drug categories, generic drugs, and drugs mentioned by their trade names. I collected Google results using a site specific search. (Example: “azathioprine” site: crohnsforum.com) The Results mentioned refer to the number of search results found by Google using the above mentioned site specific search. Quotation marks were used in each query.
I’ve been an member of crohnsforum.com since 2011. It’s a very active forum which was started in 2006. It has knowledgeable participation from experts and over 71,000 posts in the “Your Story” section. In the March 2013 newsletter the webmaster reported: “…Crohnsforum.com has seen an increase of 15,000 unique visitors PER MONTH to the site and now hosts 235,000 unique visitors monthly. That means 500 more people are visiting per day than just 3 months ago.”
Besides personal experience the list of drugs I came up with was obtained by reading patient stories, reading the titles of threads, and generally browsing through the forum. My categorization of drugs is made without any formal knowledge of pharmacology. I’m just a curious Crohn’s patient. If you have a suggestion about my drug categorization please leave a comment. Sub-categorization of pain medication into NSAIDs, opiates, etc… for example, might be useful in future analysis.
In compiling the list of words to query I was sensitive to how users of crohnsforum refer to drugs. For example azathioprine is often referred to as “aza”. It can also be referred to by its trade names, Imuran and Azasan. I’ve tried to include all renditions of the drug names but I’m sure I’ve left some out. I’ve also included the general category of drug in the table results, for example: immunosuppressant and immunosuppressive are queried along with azathioprine and their results recorded just as the individual drug names. In a few instances I explored possible common misspellings but the results were always insignificant. The convention used in the tables is that generic names, drug categories and slang usage are in lower case. Trade names are capitalized.
There are certainly many biases in studying a patient forum. Some simple ones I can think of with respect to crohnsforum include: the users are literate and speak English, the users are sometimes patient advocates (usually mothers of patients), they are well enough to use a computer, have Internet access, aren’t in complete denial, and don’t have a problem sharing their medical issues. There’s also a selection bias regarding why the patient joined, demographics like age, etc… A full list of biases could be its own post. (Again, if you’d like to add please do so in the comments section.)
The insights gained from this analysis lend a direction to asking many specific questions. Here are five: Are there any community level drug cascades? How can patient advocacy groups more efficiently allocate resources given the distribution of drugs mentioned? What are the possible dangerous drug interactions? What’s the most commonly mentioned drug side effect from each drug? How often does standard IBD treatment result in iatrogenic complaints?
In doing a peripheral search of the academic literature on data mining patient discussion boards and forums I didn’t find much. It seems like a completely wide open field. (Please make research references in the comments section.) Doing a site specific advanced Google search was a quick and dirty proof-of-concept hack. A more robust back-end infrastructure is necessary for more detailed analysis. A proper analysis would include a better drug list and the text of the forum reproduced on a dedicated server along with a scripting language with a good regular expressions library. Python or R would be the best choices.
Below are the tables used to generate the summary result table. There is one table for each category. The MISC category contains a few separate categories which should be treated individually in a more robust analysis.