Indiana’s Scott County, roughly 80 miles south of Indianapolis, experienced one of the worst opioid-driven HIV outbreaks in rural America. What happened next is a case study in data-driven outbreak intervention.
In June of 2017, the U.S. Food and Drug Administration (FDA) issued a press release which called for the removal of Opana ER (oxymorphone hydrochloride) from the market. An FDA advisory council, which voted 18-8 in favor of the drug’s removal, determined that Opana ER’s risks outweighed its benefits, and that the drug posed too great a threat to public health. The decision was based on an investigation which revealed that intravenous injection of Opana ER, an extended release opioid painkiller, led to a major HIV outbreak which spread throughout rural Scott County, Indiana beginning in early 2015. The council reached its decision thanks in part to insight derived from data analysis by the Centers for Disease Control and Prevention (CDC), which was supported by Leidos data science expertise and technology.
Endo Pharmaceuticals reformulated Opana ER in 2012 to prevent users from snorting the drug to get high. But the reformulation had unintended consequences. Users quickly discovered how to crush and inject the drug instead. This put them at a high risk for bloodborne diseases which can stay on used needles and other injection equipment. Because this equipment is often shared among people who inject drugs, outbreaks can occur at alarming rates.
This is exactly what happened in Scott County, where Opana ER’s reformulation led to an HIV problem on top of an opioid addiction problem, rightly described as an epidemic within an epidemic. The outbreak, characterized as quick and vicious, led to more than 200 confirmed cases of HIV among a population of roughly 4,500 people. Roughly 90 percent of those infected also tested positive for Hepatitis C, and many were diagnosed with thrombotic microangiopathy.
Indiana's Scott County is located roughly 80 miles south of Indianapolis.
Local health officials requested support to help respond to these alarming spikes, and invited CDC to monitor the outbreak and determine how to stop it. At the time, a team of Leidos data scientists was working with CDC to determine whether or not the company’s Collaborative Advanced Analytics & Data Sharing (CAADS) platform, a Hadoop-based big data software, could uncover undetected outbreak factors. The platform, designed to analyze large, disparate data sources in a collaborative manner, was able to not only provide valuable insight, but also reduce analysis time by a factor of six while improving interaction between data scientists and CDC officials.
The CDC sought to accomplish high-resolution epidemiological tracing, using data analytics to paint a more complete picture of the outbreak. “There were a lot of things that needed to be understood,” said Dr. Ryan Weil, Leidos Chief Scientist, “including outbreak clusters, geographic factors, epidemiological patterns, and drug resistance data.” Hidden within those data sets, he explained, were even more variables to consider as potential causal factors, including transmission through sexual encounters, needle sharing, and other data points that are not easily correlated. To complicate matters further, the data came from many different sources, including internal data, data from the field, and publicly available information.
The CDC, supported by the Leidos team, used machine learning to examine the correlation between needle sharing, sexual contact, and the contraction of HIV. Machine learning tools also predicted unexpected variables, which allowed the team to assess a person’s risk profile quickly. “This was powerful insight,” Weil said, “because while the correlation between needle sharing, sexual partners, and HIV might seem obvious, the data proved that sexual contact played an enormous role in the onward transmission in what was largely an injection drug driven outbreak.”
Most importantly, the analysis performed by CDC helped prove 90 percent of the infected population was injecting Opana ER. This analysis was used as evidence in the FDA’s decision to remove the drug from the market. It also alerted public health decision-makers to early warning factors for this specific type of outbreak. Armed with this insight, CDC published a county-by-county assessment of where similar outbreaks are likely to occur in the future.
The CDC team supported by Leidos personnel and the CAADS platform was able to understand the transmission dynamics of the Scott county outbreak. “We believe in empowering the people who understand the data and have to live with decisions made based on its insights,” Weil said. “We have people who understand data and public health making these discoveries. We want subject matter experts to be hands-on with the data. Using the platform requires no coding skills,” he said. “You don’t need a masters in statistics to use it. You can bring in new types of data and be working on it within a matter of hours. It’s very good in its ability to do quick analysis in fluid situations.”
With the ongoing opioid crisis and the role of prescription drugs in the epidemic, health officials are taking decisive action, informed by data, to help save lives. Leidos data science expertise and CAADS platform can help reduce new cases of opioid addiction and stem misuse and abuse by quickly translating disparate but meaningful data into actionable insight to inform regulatory decision-making.
Brandon is a writer and content marketer based in the Washington, D.C. area. He loves to cover emerging technology and its power to improve society.Follow on Twitter More Content by Brandon Buckner