Monitoring the opioid epidemic via social media discussions

  1. Delaney A. Smith
  2. Adam Lavertu
  3. Aadesh Salecha
  4. Tymor Hamamsy
  5. Keith Humphreys
  6. Mathew V. Kiang
  7. Russ B. Altman
  8. Johannes C. Eichstaedt

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Opioid-involved overdose deaths have risen significantly since 1999 with over 80,000 deaths annually since 2021, primarily driven by synthetic opioids, like fentanyl. Responding to the rapidly changing opioid crisis requires reliable and timely information. One possible source of such data is the social media platforms with billions of user-generated posts, a fraction of which are about drug use. We therefore assessed the utility of Reddit data for surveillance of the opioid epidemic, covering prescription, heroin, and synthetic drugs (as of September 2024, up-to-date Reddit data was still accessible on the open web). Specifically, we built a natural language processing pipeline to identify opioid-related comments and created a cohort of 1,689,039 geo-located Reddit users, each assigned to a state. We followed these users from 2010 through 2022, measured their opioid-related posting activity over time, and compared this posting activity against CDC overdose and National Forensic Laboratory Information System (NFLIS) drug report rates. To simulate the real-world prediction of synthetic drug overdose rates, we added near real-time Reddit data to a model relying on CDC mortality data with a typical 6-month reporting lag and found that Reddit data significantly improved prediction accuracy. We observed drastic, largely unpredictable changes in both Reddit and overdose patterns during the COVID-19 pandemic. Reddit discussions covered a wide variety of drug types that are currently missed by official reporting. This work suggests that social media can help identify and monitor known and emerging drug epidemics and that this data is a public health “common good” to which researchers should continue to have access.

Significance statement

The opioid epidemic persists in the United States with over 80,000 deaths annually since 2021, primarily driven by synthetic opioids like fentanyl. As the geographic and demographic patterns of the opioid epidemic are rapidly changing, accurate and timely monitoring is needed. In this paper, we used social media data from Reddit to conduct public health surveillance of the opioid epidemic, following 1.5+ million geo-located users over 10+ years. We also found that near real-time Reddit data can improve our ability to predict future overdose death rates compared to models only using CDC data with typical half-year reporting delays. Our work suggests that social media can be a useful component for public health surveillance of the opioid epidemic.

Article activity feed

  1. Version published to 10.1101/2021.04.01.21254815v2 on medRxiv
  2. ScreenIT

    SciScore for 10.1101/2021.04.01.21254815: (What is this?)

    Please note, not all rigor criteria are appropriate for all manuscripts.

    Table 1: Rigor

    Institutional Review Board Statementnot detected.
    Randomizationnot detected.
    Blindingnot detected.
    Power Analysisnot detected.
    Sex as a biological variablenot detected.

    Table 2: Resources

    No key resources detected.

    Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).

    Results from LimitationRecognizer: We detected the following sentences addressing limitations in the study:
    A key limitation of this study is that we do not distinguish between drug mentions and probable drug use, as our calculation of drug comment rates uses the number of mentions of a drug, not the number of drug mentions indicative of drug use. Future methods could make use of methods that have been developed to differentiate abuse from discussions, potentially improving accuracy of the overall system33. It is also possible that since we are not selecting for only comments regarding drug usage, that the rates of drug discussions we observe are driven by news cycles. While there is certainly an interplay between the news coverage and online discussions of drugs, the prevalence of kratom discussion on Reddit indicates that discourse can evolve without mainstream news coverage, as kratom is a drug with very low media coverage but high rates of discussion. Additionally, during the COVID-19 pandemic of the past year, changes in opioid overdose rates have received relatively little press coverage. Therefore, we view it as unlikely that news coverage drove the observed recent changes in opioid discussions or the emergence of kratom related activity. There is potential for interplay and response from the drug abuse community to the existence of a future social media surveillance system. If the drug abuse community sought to avoid surveillance efforts, those communities could find or create new platforms in which to converse, which could potentially facilitate private discussions. Indeed...

    Results from TrialIdentifier: No clinical trial numbers were referenced.

    Results from Barzooka: We did not find any issues relating to the usage of bar graphs.

    Results from JetFighter: We did not find any issues relating to colormaps.

    Results from rtransparent:
    • Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
    • Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
    • No protocol registration statement was detected.

    About SciScore

    SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

  3. Version published to 10.1101/2021.04.01.21254815v1 on medRxiv