Researchers say algorithms like NarxCare, which are used to identify a patient's risk of misusing opioids, may be actually worsening the US opioid crisis
and it's destroying lives. https://www.wired.com/... Dr. Darren Abbey / @thebiologistisn : Machine learning without validation is at best useless and will actively harm people when it is used to guide decisions about people. https://twitter.com/... @national_pain : How does Appriss' Narxcare score affect patients in real time? The latest from @maiasz explains the difficulties with algorithms dictating patient care: https://twitter.com/... Lauren O'Neal / @laureneoneal : 1. Let the Sackler family create an opioid empire worth billions 2. Create a cottage industry to surveil and punish anyone who uses opioids in any context 3. ??? 4. Profit (for rich people, hell world for everyone else) https://twitter.com/... @jurijuri : I had to stop reading this article because it was about to give me a panic attack. It really is terrifying and I recommend all you healthy folks out there read it. (Most if you chronically ill folks already know this stuff.) https://twitter.com/... Lumpy Louise / @lumpylouish : I opted to come off opioids and use cannabis because it worked for me but so many do need them to get through the day. Kathryn was kicked off them because of Narxcare deemed she was doctor “shopping” because she had dogs who were on medications too. https://twitter.com/... Shannon Okey / @knitgrrl : This article is enraging. Also it reminds me of one reason I dropped my last doctor — her smartass remarks about how (not) often I'd been refilling my Adderall prescription, therefore I didn't really need it. I could see the screen she was looking at with my drug refill report. https://twitter.com/... @wanderindreamr : “US Department of Justice has poured hundreds of millions of dollars into developing and maintaining state-level prescription drug databases—electronic registries that track scripts for certain controlled substances in real time” EW https://twitter.com/... @annaleen : A terrifying story from @maiasz about the algorithms that are denying people their pain meds — without recourse, and without reason. https://www.wired.com/... Ally / @allytncat : I have diagnosed complex PTSD due to the years of being denied adequate pain care. I haven't had lasting pain since 2016, & I am STILL unpacking the emotional trauma of it all—5 years later. Of course patient in this article has #endometriosis. Of course. Spoiler alert: I do too. https://twitter.com/... Maia Szalavitz / @maiasz : at last... my article on how trauma survivors are being denied pain care bc algorithms... https://www.wired.com/... Alex Howard / @digiphile : “nearly all Americans have the equivalent of a secret credit score that rates the risk of prescribing controlled substances to them. And doctors have authorities looking over their shoulders as they weigh their own responses to those scores.” https://www.wired.com/...
Context & Ripple Effects
The NarxCare reporting lands in a documented pattern of health algorithms harming the patients they score. Arkansas automated assessments were already cutting essential home care for disabled patients back in 2018 (algorithm-driven care cuts), and a 2020 investigation found care-targeting software infusing racial bias into decisions about who gets stepped-up treatment.
What makes this installment sharper: NarxCare sits on state prescription drug databases that the DOJ has funded with hundreds of millions, so an unvalidated score is wired directly into prescribing decisions. That extends the arc from hospitals quietly deploying unproven AI decision support into the most contested prescribing area in American medicine — with patients and prescribers as the immediate exposure.
First-order effects
- Patients flagged as high-risk by Appriss' NarxCare face denial or tapering of pain medication based on a score most don't know exists, turning a risk tool into a de facto gatekeeper at the pharmacy and clinic.
- Prescribers and pharmacists are pushed to defer to the algorithm's output over clinical judgment, echoing clinicians in related reporting who felt pressure to defer to flawed AI diagnosis tools.
Second-order effects
- Appriss faces pressure to validate NarxCare and disclose how scores affect care in real time — the same accountability demand that followed the Arkansas and racial-bias investigations, now backed by researchers rather than just journalists.
- State agencies operating the PDMPs that feed NarxCare must decide whether to keep bundling risk scores into systems built with DOJ funding, or require independent validation before scores influence prescribing.
Third-order effects
- If the pattern holds across these cases — Arkansas home care, care-targeting bias, opioid scoring — the likely endpoint is a validation and disclosure regime for clinical algorithms, treating them more like regulated medical decision tools than black-box software.
- The deeper shift is that opaque risk scoring is converging across domains: the same predictive-algorithm logic documented in police patrols, prison sentences, and probation is now shaping who receives medical treatment, putting health and criminal justice on a shared accountability path.
The trend: Health-care algorithms are drifting from advisory decision support into unvalidated gatekeeping over care, building the case for mandatory validation and disclosure rules.