Increases in Online Posts About Synthetic Opioids Preceding Increases in Synthetic Opioid Death Rates: a Retrospective O

  • PDF / 259,189 Bytes
  • 3 Pages / 595.276 x 790.866 pts Page_size
  • 91 Downloads / 162 Views

DOWNLOAD

REPORT


Division of Violence Prevention, National Center for Injury Prevention and Control, U.S. Centers for Disease Control and Prevention (CDC), Atlanta, GA, USA; 2Division of Unintentional Injury Prevention, National Center for Injury Prevention and Control, U.S. Centers for Disease Control and Prevention (CDC), Atlanta, GA, USA.

J Gen Intern Med DOI: 10.1007/s11606-019-05255-5 © Society of General Internal Medicine (This is a U.S. government work and not under copyright protection in the U.S.; foreign copyright protection may apply) 2019

INTRODUCTION

Opioid overdose deaths have increased more than fivefold from 1999 to 2016, accounting for 42,249 deaths in 2016.1 One particularly challenging aspect of the opioid epidemic is that it has been marked by a rapid transition from prescription opioids to heroin to synthetic opioids. This third wave involving synthetic opioids has largely been driven by illicitly manufactured fentanyl and its analogs.2 To address challenges of quickly identifying newly emerging synthetic opioids, novel data sources such as web or social media data may serve as potential early warning systems. Prior work has mainly focused on automated identification of messages indicating misuse,3 detection of online illicit pharmacies, evaluating opinions around certain compounds,4 understanding spread of norms, and comparing online findings to survey data.5 However, there is particularly limited work examining fentanyl and fentanyl analogs (now the leading cause of overdose deaths) and limited work that directly compares findings from these novel approaches to death data to describe how much lead time an early warning system based on online data could potentially provide. Thus, this retrospective analysis sought to assess the degree to which such an early warning system could have provided early insights about the rise in synthetic opioids deaths.

1, 2010, through December 31, 2017, which were queried from a public repository of such data (https://pushshift.io/) using the site’s Application Programming Interface. Our primary objective was to assess trends in posts about synthetic opioids, primarily fentanyl and fentanyl analogs. Identification of such messages is challenging as multiple fentanyl-related compounds exist and the presence of drug misspellings and vernacular is prevalent. Consequently, we used two open-source algorithms that cluster words based on semantic similarity, word2vec and fastText, released by researchers at Google and Facebook, respectively. Such algorithms can be used to detect misspellings of a word (e.g., fentanyl and fentanil) as well as words that are similar in meaning (e.g., syringe and needle). We used these algorithms to identify a list of 34 keywords (Fig. 1) that captured the leading terms in synthetic opioid posts and plotted volume of synthetic opioid posts over time. Our examination of anonymous, publicly available posts was granted IRB exemption.

RESULTS

A total of 27,210,148 messages (posts or comments) were assessed. Figure 1 reveals that the rate of online posts a