Paracelsus once said “All things are poison, and nothing is without poison; only the dose permits something not to be poisonous.”

This has often been shortened and quoted simply as “The dose makes the poison.” [1] As a pharmacist, I can tell you that any medication is likely to have a side effect with differing levels of severity. Sometimes these side effects are desirable, such as opioids antitussive effects. Too much, however, could lead to an adverse event such as respiratory depression.

Identifying adverse effects is critical and now a group of researchers is proposing leveraging the power of internet searches to find them faster.

Identifying these drug side effects is often the most difficult part. When testing a new medication in clinical studies, adverse effects are noted and identified. Studies involving new drugs often include a table or figure dedicated to identifying the common side effects discovered. These are then included in the drug monograph once the drug is approved by the FDA. Finally, they then get inserted into the minuscule print of patient education printouts that patients receive when they pick up the prescription.

Needless to say, some medication side effects that are under-reported or not initially identified become better understood when the general population starts to use these medications routinely. Particularly for those effects that may be extremely rare, its often the general practitioner who first identifies the connection.

The FDA requires a report when a healthcare professional feels a medication has caused an event. Unfortunately, the inherent limitation here is that it requires a health professional to identify the adverse effect as related to the drug – and the patient often has no direct input unless they reach out themselves.

Some drug companies do not like this requirement. While patients could easily go and call the company and alert them to the event, the rise of social media has caused new issue. Case in point – on August 15, 2011, Facebook stopped Pages from blocking comments. For pharmaceutical companies that had pages set up for their medications, this was a problem. If patients started listing their side effects or adverse events, then the pharmaceutical companies were obligated to follow-up with those patients and then report the results to the FDA. Many companies simply shut down their pages, as well as other social media outlets, that gave patients the opportunity to identify these issues.

This is a pity. The fact remains that patients are more likely to identify a side effect or event first and before their practitioner. In an age of big data, how can we use patients to create a better surveillance system to help identify the risk?

Recently a study published by White and colleagues assessed whether collecting and analyzing anonymized data collected via internet searches could correlate drug interactions. [2] The authors particularly looked at the interaction between pravastatin and paroxetine. The interaction between these two drugs is that blood glucose elevations may occur with the combination. [3]

The investigators looked at millions of users’ (who had a web browser add-on installed with consent) internet searches involving pravastatin and paroxetine and terms affiliated with hyperglycemia (e.g. blurry vision, fatigue). Interestingly, the data demonstrated a correlation of those on the two medications with search activities for terms associated with hyperglycemia.

This is a great idea to utilize web-browser history to look into possible side effects and drug interactions by users. As mentioned earlier, it bypasses the need for others to do the reporting and offers an alternative method to find data and investigate possible issues. These methods could also be expanded to the use of social media, especially with Facebook Graph coming out shortly. Nonetheless, patient information needs to be safeguarded and will prove to be the biggest barrier to this method.


1. p. 435, Verkehrsmedizin: Fahreignung, Fahrsicherheit, Unfallrekonstruktion, B. Madea, F. Mußhoff, and G. Berghaus, Köln: Deutscher Ärzte-Verlag, 2007 (Identified via Wikipedia)
2. White RW, Tatonetti NP, Shah NH, Altman RB, Horvitz E. Web-scale pharmacovigilance: listening to signals from the crowd. J Am Med Inform Assoc. 2013;
3. Tatonetti NP, Denny JC, Murphy SN, et al. Detecting drug interactions from adverse-event reports: interaction between paroxetine and pravastatin increases blood glucose levels. Clin Pharmacol Ther. 2011;90(1):133-42.