Subscribe now

Health

Side effects kill thousands but our data on them is flawed

By Mallory Locklear

10 August 2017

New Scientist Default Image

Kill or cure? Some drugs have unwanted side effects

Skhoward/Getty

Side effects are a serious problem. As many as 40,000 people in the US die from drug side effects a year. To tackle the problem, the US Food and Drug Administration collects data from people who experience adverse events while taking prescribed drugs, but an analysis of that database has now discovered that it is full of noise and errors.

The Food and Drug Administration Adverse Event Reporting System (FAERS) works in a similar way to the UK’s Yellow Card Scheme, enabling anyone to report drug side effects, not just doctors. Because the database is continuously updated, it’s often used by researchers seeking to understand and reduce drug side effects.

But when one team attempted to use the database for a study like this, they found the data too flawed to use. This prompted Brian Shoichet of the University of California, San Francisco, and colleagues at Novartis Institutes of Biomedical Research, to take a closer look.

Analysing the data in FAERS, they found a wide range of problems. Many reports were submitted more than once, and often, the disease being treated was wrongly listed as a side effect of the drug – diabetes being named as a side effect of diabetes medication, for example.

An imprecise goldmine

One drug can have several names – a chemical name, plus brand names – and the database doesn’t account for this. The team found that to get a full picture of possible side effects for a particular drug, they had to include every one of its synonyms in their database query. In some cases, this meant including hundreds of names. Shoichet says he thinks these problems probably have a huge impact on drug research.

“Some of these issues must be addressed through careful data cleansing, and others through analysis,” says Rave Harpaz, a researcher at technology firm Oracle Health Sciences in California. He describes the database as a tradeoff between large amounts of data versus precision.

“In principle, it’s a goldmine,” said Shoichet who along with his colleagues suggested a few ways that systems like FAERS can be more useful, such as automatically linking drug names to their ingredients and requiring information like dosage.

Savannah Irving, a doctor in the UK, says from a research perspective, databases like FAERS are invaluable. For example, a recent project initiated by the Chicago Tribune used FAERS data to discover a potentially fatal drug interaction. But doctors cannot rely on databases like these alone. “You’re starting a new drug that is new for that patient,” she says, meaning each person should be monitored for side effects, regardless of what is already known.

Journal reference: eLife, DOI: 10.7554/eLife.25818

Topics:

Sign up to our weekly newsletter

Receive a weekly dose of discovery in your inbox! We'll also keep you up to date with New Scientist events and special offers.

Sign up