Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Correspondence
  • Published:

Mitigating algorithmic bias in opioid risk-score modeling to ensure equitable access to pain relief

Nature Medicine volume 29, pages 769–770 (2023)Cite this article

Subjects

Prescription drug monitoring programs (PDMPs) are statewide electronic databases that track prescriptions of controlled substances by both physician prescribers and pharmacists. Opioid prescribers and pharmacists use these databases to ensure safe prescribing levels for their patients. Bamboo Health is the developer of NarxCare, an artificial intelligence (AI)- and machine learning (ML)-based algorithm implemented in 46 of the 51 US PDMPs1,2. The NarxCare algorithm draws on data from the PDMP electronic database and other sources2,3,4 (Table 1). The NarxCare algorithm distills this information into a three-digit NarxScore within a patient’s PDMP profile; the score is designed to predict their risk for opioid misuse or overdose. Clinicians and pharmacists are instructed to check a patient’s score when prescribing opioids, to prevent opioid misuse2.

Table 1 PDMP and non-PDMP data that may be available to the NarxCare algorithm
Full size table

Given the widespread implementation of NarxCare throughout the United States, ensuring that its risk predictions are equitable for all patients is essential. However, as with other ML-based algorithms, NarxCare may create biased predictions that harm marginalized and under-represented patient populations5. This is especially of concern because physicians may terminate all medical care for patients with high risk scores out of fear of prosecution by law-enforcement agencies6. Patients have described being abruptly removed from clinics during appointments due to high NarxCare scores. For example, one patient classified as high risk because her pets’ benzodiazepine prescriptions were listed under her PDMP profile was promptly removed from the clinic and had her relationship with her physician terminated7.

This is a preview of subscription content, access via your institution

Access options

Buy this article

  • Purchase on Springer Link
  • Instant access to full article PDF
Buy now

Prices may be subject to local taxes which are calculated during checkout

Acknowledgements

We thank J.M. Sharfstein for reviewing and providing feedback on this manuscript. We also thank R. Gibbons for helpful discussions on evidence-based screening tools, and M. Vogel for insight on ethics surrounding artificial intelligence and algorithms in healthcare. The views, opinions, and findings in this article are those of the authors. This project was funded through a grant from Bloomberg Philanthropies. The funding organization had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Author information

Authors and Affiliations

  1. Johns Hopkins University, Baltimore, MD, USA

    Atharva M. Bhagwat

  2. Department of Health Policy & Management, Johns Hopkins University Bloomberg School of Public Health, Baltimore, MD, USA

    Atharva M. Bhagwat, Kadija S. Ferryman & Jason B. Gibbons

  3. Berman Institute of Bioethics, Johns Hopkins University, Baltimore, MD, USA

    Kadija S. Ferryman

Authors
  1. Atharva M. Bhagwat
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Kadija S. Ferryman
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Jason B. Gibbons
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Jason B. Gibbons.

Ethics declarations

Competing interests

K.S.F. is a member of the digital ethics advisory board for Merck and of the institutional review board for the US National Institutes of Health’s All of Us Research Program. The other authors declare no competing interests.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Bhagwat, A.M., Ferryman, K.S. & Gibbons, J.B. Mitigating algorithmic bias in opioid risk-score modeling to ensure equitable access to pain relief. Nat Med 29, 769–770 (2023). https://doi.org/10.1038/s41591-023-02256-0

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s41591-023-02256-0

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing