A call for better validation of opioid overdose risk algorithms

Abstract

Clinical decision support (CDS) systems powered by predictive models have the potential to improve the accuracy and efficiency of clinical decision-making. However, without sufficient validation, these systems have the potential to mislead clinicians and harm patients. This is especially true for CDS systems used by opioid prescribers and dispensers, where a flawed prediction can directly harm patients. To prevent these harms, regulators and researchers have proposed guidance for validating predictive models and CDS systems. However, this guidance is not universally followed and is not required by law. We call on CDS developers, deployers, and users to hold these systems to higher standards of clinical and technical validation. We provide a case study on two CDS systems deployed on a national scale in the United States for predicting a patient’s risk of adverse opioid-related events: the Stratification Tool for Opioid Risk Mitigation (STORM), used by the Veterans Health Administration, and NarxCare, a commercial system.

Challenging Disability Discrimination in the Clinical Use of PDMP Algorithms

“…As explained above, PDMP risk‐scoring algorithms have not been externally validated by either peer review or the Food and Drug Administration, and researchers have raised serious questions about the validity of PDMP scoring systems. A 2023 study comparing two such systems explains that both technical validation , whether the tool accurately and reliably predicts what it aims to predict, and clinical validation , whether the tool yields its intended impact on providers and patients, are critical for preventing serious harms to patients 32 . As noted in the proposed section 1557 rule, the American Medical Association has issued similar guidance about the evaluation, use, and monitoring of artificial intelligence systems and clinical algorithmic tools in health care.…”

“…Individualized assessment is especially important given the lack of external validation of PDMP algorithms and the risk scores they generate. As noted in a 2023 study of two scoring systems, if PDMP information and associated risk scores are not sufficiently validated, they can mislead providers and directly harm patients 28 . Indeed, patients can be harmed by these things even when they are validated.…”

“…As noted in a 2023 study of two scoring systems, if PDMP information and associated risk scores are not sufficiently validated, they can mislead providers and directly harm patients. 28 Indeed, patients can be harmed by these things even when they are validated. Assessing the patient's individual circumstances and needs is a mechanism to address serious errors, omissions, and biases that find their way into PDMP algorithmic data.…”