Battling the Opioid Epidemic, Byte by Byte
Earlier this year, the FDA announced a $20 million plan to harness the power of Big Data to address the nation’s opioid crisis. Widely referred to as the nation’s biggest public health crisis, the opioid epidemic dates back to liberal opioid dispensing that began in the late 1990s.[1] Of the more than 700,000 overdose-related deaths in the United States during the period from 1999 through 2017, a staggering 68% involved opioids.[2] The total number of deaths includes those related to both prescription and illegally-acquired opioids, but the numbers shown below, supplied by the National Institute on Drug Abuse[3], indicate the latter is often a result of the former:
- Out of 100 patients prescribed opioids for chronic pain, as many as 29 end up misusing them.
- Out of those 29 people who misuse prescription opioids, as many as 6 will transition to heroin.
- Out of 100 people who use heroin, 80 first misused prescription opioids.
What’s come to light over the last two decades, however, is that many who were prescribed opioids could have managed their pain with non-opioid pain therapy (e.g., ibuprofen, acetaminophen, and physical therapy).[4] And for those at higher risk for opioid addiction, opioids are precisely the wrong therapy for pain management. But who’s at higher risk, and how can healthcare providers identify them?
That’s where Big Data analytics comes in. Why not use the vast amounts of information that exist on who’s been prescribed what, by whom, for what, and how that’s turned out, to prescribe more intelligently? Of course, the data has to first be properly recorded, structured, unified, and organized (i.e., data management), and the appropriate analytical tools need to be developed and applied (i.e., big data analytics). Here are some ways Big Data analytics is currently being brought to bear on the opioid epidemic:
PDMPs
“PDMP” stands for “prescription drug monitoring program,” which the CDC calls the “most promising state-level intervention” toward improving the practice of opioid prescribing.[5] PDMPs can provide health authorities timely information about prescribing and patient behaviors that contribute to the epidemic and facilitate a nimble and targeted response. As of April 2019, all 50 states[6] have adopted PDMP legislation (the lone holdout is Missouri), 37 of which are fully operational.
EHRs
The first electronic health record (“EHR,” which is also referred to as an electronic medical record) was developed in 1972[7]. The use of EHRs increased dramatically after the introduction of the Health Insurance Portability and Accountability Act (HIPAA) in 1996 (as a more efficient means of complying with HIPAA). By 2014, the call had been made for industry-wide adoption of EHRs. Along with their many other benefits, EHRs can effectively supplement traditional surveillance methods for monitoring trends in opioid prescribing, with minimal lag time, according to the CDC[8].
Health IT Integration
Making PDMP information available via EHR can make it easy for providers to quickly assess whether a patient should or should not be prescribed opioids, notes the Office of the National Coordinator for Health Information Technology,[9] which has been working toward PDMP and EHR integration. Specifically, the PDMP & Health IT Integration Initiative has been working toward establishing a standardized approach to delivering data stored in the PDMPs to EHRs, pharmacy systems, and health information exchanges.
The FDA’s Database Project
The FDA’s project referred to here will attempt to take the aforementioned efforts even further by using Big Data to identify not just individuals, but communities, at risk with regard to opioid addiction.
Harnessing Social Media
The next level of addiction prediction may come from mining social media. New Jersey Institute of Technology (NJIT)[10] has been developing a real-time data analytics tool to monitor social platforms such as Twitter and Reddit to help professionals identify and treat drug abuse. The tool is expected to use, among other things, machine learning (also known as artificial intelligence, or AI), to determine what online behavior correlates with drug abuse.
