Claims data will power new research initiative developing predictive models to fight opioid crisis
July 11, 2019
By Bill Lucia
Today, Americans are more likely to die from an accidental opioid overdose than a car accident, according to the National Safety Council.
And as the opioid crisis continues to ravage the U.S., it’s easy to lose sight of the fact that it is a global dilemma that extends to virtually all corners of the developed world.
Australia, for example, continues to see rising use of opioids in its population, with doctors in the country having written more than 14 million prescriptions for opioid pain killers in 2017. One in 10 Australians who are prescribed a pain killer become addicted, with serious consequences for addicts and their families. In 2015, 69 percent of Australia’s drug-related deaths were due to prescription drugs, according to AddictionCenter.
Now, the opioid crisis’ reach has prompted new action from the Australian government. The Digital Health Cooperative Research Centre (CRC), a research group funded by the Australian government, recently announced plans to collaborate with U.S.-based technology company HMS and several U.S. universities to mine claims data with the goal of developing predictive analytics models that will help clinicians better predict and solve health issues, deliver better patient outcomes and ultimately, save lives.
The research collaborative’s first target is opioid abuse.
Big data to improve outcomes
This research effort is all about the shared common goal of harnessing big data and digital technologies to deliver better outcomes in healthcare, with each member of the coalition contributing its own unique and deep expertise. The Digital Health CRC — which is the largest digital health research cooperative in the world, with funding from the Australian government, universities and businesses totaling more than $160 million — will deliver funding, expert guidance and research tools.
Our university partners, which include Stanford University and Southern Methodist University, will provide Ph.D. candidates in healthcare economics, biostatistics, data science, public health and other related fields to perform the research.
HMS, which provides analytics and engagement solutions that help reduce healthcare costs and improve outcomes, will ensure researchers have a significant volume of data. HMS serves more than 40 state Medicaid agencies, 325 health plans, 150 employers and three federal agencies, so we have large volumes of paid claims data that can be de-identified for research purposes. The enormous size and scope of our centralized data set is a critical element of the project.
Why claims data?
Healthcare in Australia — like the U.S. — is fragmented across many different siloes, including general practice (primary care), specialist care, hospital care, geriatric care and others. This fragmentation of care also makes it extremely difficult to access the type of comprehensive and cohesive health data from patient records that is necessary to underpin strong research.
While other researchers are already doing work with electronic health records (EHRs) and clinical data, we believe that paid claims data represent a different and unique — and more importantly, underutilized — data set.
For example, one patient may have multiple records scattered across different EHR systems, such as her primary care provider, any specialist she sees and her hospital. In contrast, a paid claims data set aggregates a patient’s entire medical claims history — physician, specialist, pharmacy, hospital, even dental and vision — in one place.
The paid claims data sets we work with are longitudinal and span many years, which is critical for developing predictive models. Further, the fact that paid claims are in standard electronic formats allows us to run the same algorithms across multiple data sets to compare results and determine what may be regional versus universal.
Identifying risk factors for opioid misuse
The first initiative of the research project will be led by Stanford University and will focus on identifying potential risk factors for opioid misuse. Specifically, Stanford researchers will endeavor to develop predictive models that identify three things: 1) opioid misuse disorder risk factors for patients who have no previous abuse history; 2) providers who have risky opioid prescribing patterns; and 3) patients at risk of relapsing into opioid misuse.
We anticipate having initial research results to share from this portion of the opioid project as early as the end of this August.
The second research initiative will be led by Southern Methodist University and will center on developing a predictive analytics model to reduce the rate of hospital readmissions. Hospital readmissions sometimes result from suboptimal patient care, and represent an opportunity to lower costs, improve quality, and increase patient satisfaction all at once.
The Southern Methodist University-led initiative will develop a predictive model to help clinicians identify patients who are more likely to be re-admitted, enabling clinicians to engage with these patients with the objective of preventing future readmissions.
Eventually, the algorithms developed by these researchers could be commercialized and included in population health management systems to help clinicians around the world deliver better care.
It’s time that Americans realize the opioid crisis stretches far beyond our own borders. Across the world, we are seeing upticks in cost, readmission and opioid use. It is imperative that as an industry we act together to solve it — globally.
About the author: Bill Lucia is the CEO of HMS, a technology, analytics and engagement solutions company that provides a broad range of coordination of benefits, payment integrity, care management and member engagement solutions to help move the healthcare system forward.