Analysis of doc notes in EHRs can flag drug interactions

Stanford University researchers touted success in using analysis of free-text notes in electronic health records for surveillance of drug interactions in near real time.

Their work, published online this week in Clinical Pharmacology & Therapeutics, noted that not all safety issues with new drugs are identified before they hit the market. Their approach, they said, can reduce the time that patients are at risk--providing more timely alerts than previous research based on discharge summaries or insurance claims. Reactions can be identified even before an official alert goes out, the authors said.

Their processing method automatically removes identifying patient information, removing one of the prevailing barriers to using this data.

From their data set, they noted 28 positive associations and 165 negative associations covering 78 drugs, and 12 events related to a single drug. For the drug-drug interactions, the reference set contained 466 positive and 466 negative associations spanning 333 drugs across 10 events.

The authors said the evolution of better tools in natural language processing will help speed up the process.

Similar research from Vanderbilt University published last November looked at lab results for patients given a particular drug to a matched group who did not receive the drug. That paper called EHRs a "valuable resource" to improve medication safety by detecting previously reported adverse drug reactions and detecting new ones.

Patients' Internet search patterns can be another useful tool in flagging drug reactions, according to research from Microsoft, Stanford and Columbia University. The researchers combed through logs of patient queries entered into Google, Microsoft and Yahoo search engines to look for side effects from taking a combo of the antidepressant paroxetine and the cholesterol-lowering drug pravastatin. The pairing later was reported to cause hyperglycemia, but researchers found a "strong signal" to indicate that in the search data.

"There is a potential public health benefit in listening to such signals and integrating them with other sources of information," the authors wrote.

Meanwhile, Stanford University researchers created an algorithm to allow physicians to determine whether a reaction was drug-related, or due to some other illness.

To learn more:
- find the study