Cutting through the chatter: Toxicovigilance and Twitter

toxicoviligance and Twitter_

Tracking the rate and incidence of nonmedical use of prescription medications has typically been done by agencies like the U.S. Food and Drug Administration through its MedWatch program or through data from the National Poisoning Data System. However, researchers from Arizona State University and the University of Arizona have found a new, unlikely source to monitor toxicovigilance: Twitter.

Their analysis showed that Twitter can be a valuable resource for signals of medication abuse-related information and proved that implementing a natural language processing system can be an important tool in monitoring and intervening.

Researchers collected tweets associated with three commonly abused medications — Adderall, oxycodone and quetiapine — as well as a control medication (metformin), which is commonly used but not as susceptible to abuse. They developed language-based signals to signify abuse and designed a classification technique to distinguish between tweets that had clear signals of abuse and those that didn’t. Their classification approach achieved 82 percent accuracy.

Results showed that there were different monthly usage patterns for Adderall and oxycodone. The number of tweets related to Adderall abuse was high during exam times, for example. However, tweets for both medications peaked around the holiday season.

The process of annotating a total of 6,400 tweets was laborious, and the intent to abuse prescription medication was often difficult to determine because of the character limitations on Twitter. However, researchers were able to classify tweets as abuse-related if they didn’t mention clinical diagnoses but rather focused on potential non-medical benefits.

One abuse-indicating tweet, for example, read: “popped Adderall tonight hahahah let’s finish this 100 page paper.” Another read: “an oxycodone high from snorting lasts for one hour. If it is swallowed, your [sic] looking at a three hour high.”

Social listening has been proven to be an effective way of monitoring misuse of drugs. Earlier this year, researchers from several pharmaceutical companies sifted through posts on two harm-reduction forums to find posts indicating misuse of… (continue reading)