Machine Learning May Help Predict Success of Prescription Opioid Regulations
Hundreds of laws aimed at reducing inappropriate prescription opioid dispensing have been implemented in the United States, yet due to the complexity of the overlapping programs, it has been difficult to evaluate their impact. A new study by researchers at Columbia University Mailman School of Public Health uses machine learning to evaluate the laws and their relation to prescription opioid dispensing patterns. They found that the presence of prescription drug monitoring programs (PDMPs) that give prescribers and dispensers access to patient data were linked to high-dispensing and high-dose dispensing counties. The findings are published in the journal Epidemiology.
“The aim of our study was to identify individual and prescription opioid-related law provision combinations that were most predictive of high opioid dispensing and high-dose opioid dispensing in U.S. counties,” said Silvia Martins, MD, PhD, associate professor of epidemiology at Columbia Mailman School. “Our results showed that not all prescription drug monitoring programs laws are created equal or influence effectiveness, and there is a critical need for better evidence on how law variations might affect opioid-related outcomes. We found that a machine learning approach could help to identify what determines a successful prescription opioid dispensing model.”
Using 162 prescription opioid law provisions capturing prescription drug monitoring program access, reporting and administration features, pain management clinic provisions, and prescription opioid limits, the researchers examined various approaches and models to attempt to identify laws most predictive of county-level and high-dose dispensing in different overdose epidemic phases—the prescription opioid phase (2006-2009), the heroin phase (2010-2012), and the fentanyl phase (2013-2016)—to further explore pattern shifts over time.
PDMP patient data access provisions most consistently predicted high-dispensing and high-dose dispensing counties. Pain management clinic-related provisions did not generally predict dispensing measures in the prescription opioid phase but became more discriminant of high dispensing and high-dose dispensing counties over time, especially in the fentanyl period.
“While further research employing diverse study designs is needed to better understand how opioid laws generally, and specifically, can limit inappropriate opioid prescribing and dispensing to reduce opioid-related harms, we feel strongly that the results of our machine learning approach to identify salient law provisions and combinations associated with dispensing rates will be key for testing which law provisions and combinations of law provision work best in future research,” noted Martins.
The researchers observe that there are at least two major challenges to evaluating the impacts of prescription opioid laws on opioid dispensing. First, U.S. states often adopt widely different versions of the same general type of law, making it particularly important to examine the specific provisions that make these laws more or less effective in regards to opioid-related harms. Second, states tend to enact multiple law types simultaneously, making it difficult to isolate the effect of any one law or specific provisions.
“Machine learning methods are increasingly being applied to similar high-dimensional data problems, and may offer a complementary approach to other forms of policy analysis including as a screening tool to identify policies and law provision interactions that require further attention,” said Martins.
Co-authors are Emilie Bruzelius, Jeanette Stingone, Hanane Akbarnejad, Christine Mauro, Megan Marzial, Kara Rudolph, Katherine Keyes, and Deborah Hasin, Columbia University Mailman School; Katherine Wheeler-Martin and Magdalena Cerdá, NYU Grossman School of Medicine; Stephen Crystal and Hillary Samples, Rutgers University; and Corey Davis, Network for Public Health Law.
The study was supported by the National Institute on Drug Abuse grants DA048572, DA047347, DA048860 and DA049950; the Agency for Healthcare Quality and Research, grant R18 HS023258; and the National Center for Advancing Translational Sciences and the New Jersey Health Foundation, grant TR003017.