Editorial Note: This post is by Ken Spriggs, who wrote the second ever RxISK story – on Azathiopine. This should be revisited in the light of the AbbVie Campaign. The current post is taken from his ongoing blog on Crohn’s Disease – a condition that is high on the target list for the makers of biologicals including AbbVie and Humira.
Drug side effect ratios explained
This is the second part in a series of research articles focused on drugs discussed by IBD patients on crohnsforum.com. The concept of Drug Side Effect Ratio is defined.
The previous article took a look at what drugs IBD patients discuss on crohnsforum. This article examines the instances of the phrase “side effects” near the drug names. I collected Google results by using a site specific search then determined the percentage of mentions of “side effect” near the drug name.
Here’s an example, Google search: [“Remicade” “side effects” site:crohnsforum.com]. There were 6,250 results. From the previous article the total mentions of Remicade is 22,100. Dividing side effects results by total results we get 0.28. So the result is that 28% of the mentions of “Remicade” occur close to the phrase “side effects”. I’ll call this metric the Drug Side Effect Ratio. The higher the ratio the more we expect the discussions in the forum are focused on side effects.
I’ve done the analysis on a subset of the drugs used in the previous article. I’ve picked the top five common drug categories and only the drugs in each category which have the highest number of mentions. I also treated the drug category ‘biologic’ as a specific drug name. Generics are in lower case and trade names are capitalized.
Let’s do a quick sanity check on the results. Notice that all the ratios are less than one. That’s good because if they weren’t it would mean there are more mentions of “side effects” than the drug name. Now consider the ranges of the distribution. The highest Drug Side Effect Ratio is for natalizumab which was withdrawn by the manufacturer and subsequently returned under the special prescription program. It’s not approved in the EU for Crohn’s treatment specifically because of side effects. Natalizumab’s trade name, Tysabri, has the second highest Drug Side Effect Ratio. The lowest Drug Side Effect Ratio is for metronidazole, but isn’t remarkably lower than the median. It’s trade name, Flagyl, on the other hand, is above the mean. It also seems reasonable to me that 76% of the discussion about biologics concerns itself with side effects.
Since biologics has the highest ratio let’s take a closer look at that category.
Drug side effect ratios for biologics
There’s a trend which seems to emerge. The trade names and generic names for the same drug have considerably different Drug Side Effect Ratios. Are people biased to use the generic name when discussing side effects? Does advertising play a role in the mentions of a trade name when mentioning side effects? It appears as if this is the trend when patients are discussing biologics. The observation doesn’t hold true for metronidazole (Drug Side Effect Ratio = 25%) and it’s trade name Flagyl (Drug Side Effect Ratio = 56%). What’s the context of discussion when a trade name is used versus a generic? Could sentiment analysis using natural language processing tell us anything about how patient dialogue differs regarding generic names and trade names?
More work is needed but it looks plausible that the Drug Side Effect Ratio could be useful in analyzing a patient message board like crohnsforum. For investigating the Drug Side Effect Ratio further a more rigorous parsing of the text is necessary. It’s possible the technique could be extended to using specific side effects near the drug names. For example the term fatigue could be counted near the drug name azathioprine.
Perhaps the results don’t say much in isolation but I feel like this analysis sheds light on what’s important in the IBD community. Drug Side Effect Ratio is certainly a step in considering descriptive generalized metrics about online patient communication.