As we reported on Vector last year, once a new drug is on the market, regulators rely on a mix of surveillance, reporting and data mining to detect adverse drug events (aka side effects).
While those methods can work pretty well, a team of scientists from Microsoft, Stanford and Columbia wanted a better way to find rare or unforeseen interactions between drugs and recently tried a new tactic: looking at what people type into Internet search engines like Google, Microsoft and Yahoo. They hit pay dirt, unearthing evidence that the combination of two drugs—the antidepressant paroxetine and a cholesterol-lowering drug called pravastatin—leads to high blood sugar.
“Given how often patients turn to the Internet for information about the drugs they are taking, it’s not unexpected that we will identify new side effects sooner,” says John Brownstein, PhD, leader of the Computational Epidemiology Group in the Children’s Hospital Informatics Program (CHIP), and whose MedWatcher mobile app takes a crowdsourcing approach to drug side-effect reporting.
The work is another demonstration of the power that search tools, social media and other alternative data sources can bring to public health surveillance. In 2011, Brownstein and colleagues demonstrated that Google searches could reveal a lot about peoples’ health behaviors. “There is tremendous promise in a wide range of tools, from online search to patient forums. We are just now at the start of a new era for drug safety surveillance,” Brownstein notes.
At the same time, the work also emphasizes the need for the public, Internet companies, privacy advocates, health care thought leaders and other stakeholders to agree on ground rules for using data like these for health surveillance. We are, after all, in an era in which everything we do is online.
“As we uncover new uses for these data, there is an important conversation to be had,” says Ben Reis, PhD, leader of CHIP’s Predictive Medicine Group. Reis is working on ways of mathematically predicting possible adverse events. “We have to ensure that the public understands both the potential value of the data for helping society at large, as well as the safeguards that are in place to protect individual privacy.”