One common use of the internet is to search for health information. Could an analysis of search terms, and a little detective work, help provide early clues about drug side effects, before more traditional methods such as official reporting systems have kicked in?
This study examined the search logs of millions of US web users who had agreed to share their searches, anonymously and automatically, with Microsoft, during 2010. The authors particularly focussed on an interaction between two drugs (paroxetine and pravastatin) causing hyperglycaemia (high blood sugar) – an association only identified in 2011. So they tested whether people using both these drug names as search terms in 2010 – when they could not have known about the interaction – were more likely to include a word associated with hyperglycaemia than those searching for just one of the medicines.
The answer was yes: about 10% of those who searched for both drugs also performed hyperglycaemia-related searches, compared with 4% or 5% from the groups who just searched for either paroxetine or pravastatin. The analysis also showed that prescription drug-related searches were common, with 1 in 250 users looking up at least one of the top 100 drugs in the USA during the year. It also looked at 31 other pairs of drugs with a known association with high blood sugar (without such convenient timing as paroxetine and pravastatin, but where the information had been published in highly technical or otherwise inaccessible outlets), and with another 31 pairs associated with a randomly chosen side effect (with any real associations removed). It was possible to identify most of the known associations from the search logs.
The authors see a “potentially valuable signal” in these search logs, and liken the system to a large network of sensors for identifying possible side effects. Early days for a new early warning system perhaps – and a fascinating use of internet search data.
Tatonetti NP, Denny JC, Murphy SN, Fernald GH, Krishnan G, Castro V, Yue P, Tsao PS, Kohane I, Roden DM and Altman RB. Detecting drug interactions from adverse-event reports: interaction between paroxetine and pravastatin increases blood glucose levels. Clin Pharmacol Ther http://dx.doi.org/10.1038/clpt.2011.83.
White RW, Tatonetti NP, Shah NH, et al. J Am Med Inform Assoc Published Online First 6 March 2013 http://jamia.bmj.com/content/early/2013/02/05/amiajnl-2012-001482.full.