« Back to Results
Applying Regulation, Medication and Big Data to the Opioid Crisis
Sunday, Jan. 6, 2019
8:00 AM - 10:00 AM
Health Economics Research Organization & American Economic Association
University of Southern California
Using Big Data and Data Science to Generate Solutions to the Opioid Crisis
The opioid epidemic is a national public health emergency. Using machine learning techniques and secure, anonymized administrative data, we are able to identify patients at risk of negative outcomes before their first opioid prescription with a high degree of accuracy. Our predictive model assigns an individual risk score that measures how likely a patient is to have a negative outcome, such as dependency, abuse or poisoning. Individual risk scores may be used to aid providers in making better, data informed decisions when weighing the medical benefit of opioid therapy against the risk of negative outcomes. Our approach puts administrative data to work to help governments curb the opioid crisis.
Medication-Assisted Treatment of Opioid Use Disorder Among Commercially Insured Individuals
As the opioid crisis continues, policymakers and insurers (particularly commercial insurers) have devoted little systematic effort to identifying optimal treatment; rather, they have focused on the important issue of prevention. However, given the high number of commercially-insured individuals with active opioid use disorder in the U.S. today (approximately 622,000 commercially-insured individuals in 2015 were diagnosed with OUD), identifying effective and affordable treatment pathways is essential. Physicians specializing in treatment of opioid use disorder have advocated the expansion of medication-assisted treatment (MAT) as a potential solution. These medications are both relatively low-cost and have been shown in clinical studies to reduce use and improve retention over other treatments and placebo drugs. However, their effect in a large-scale setting have not been evaluated. Until 2002, patients receiving MAT generally received methadone, which must be dispensed daily at one of about 1,000 federally certified centers nationwide. In 2002, buprenorphine was approved as a form of MAT that could be prescribed for take-home use by physicians who obtained a DEA waiver. However, take-up by physicians was slow and remains low, with fewer than 5000 waivered physicians in 2006, and roughly 25000 in 2017. Using a large sample of commercially insured patients over the period 2008-2016, we study the effect of access to buprenorphine on total health spending for those with opioid-use disorder. We find that total healthcare spending for patients with opioid use disorder is lower in counties with a higher share of physicians approved to prescribe buprenorphine (in models including state-year fixed effects; we are currently working to gather time-varying county-level data on buprenorphine-approved physicians so as to estimate models with county fixed effects). We decompose these decreases across spending categories, and also look at the effect on clinical outcomes (e.g. overdose incidence) to better understand the welfare
Health Insurance, Price Changes, and the Demand for Pain Relief Drugs: Evidence from Medicare Part D
Overdose deaths from prescription opioids are on the rise, and policymakers seek solutions to curb opioid misuse. Recent proposals call for price-based solutions, such as opioid taxes and removal of opioids from insurance formularies. However, there is limited evidence on how opioid consumption responds to price stimuli. This study addresses that gap by estimating the effects of prices on the utilization of opioids as well as other prescription painkillers. I use nationally representative individual-level data on prescription drug purchases to exploit the introduction of Medicare Part D in 2006 as an exogenous change in out-of-pocket drug prices. I find that new users have a relatively high price elasticity of demand for prescription opioids, and that consumers treat over-the-counter painkillers as substitutes for prescription painkillers. My results suggest that increasing out-of-pocket prices of opioids, through formulary design or taxes, may be effective in reducing new opioid use.
University of Georgia
Arizona State University