Despite a tremendous increase in awareness and a number of programs in place to address drug use, drug overdose deaths continue to rise, according to data from the Centers for Disease Control and Prevention. From 2015 to 2021, drug overdose deaths nearly doubled. This increase has been largely driven by a rising number of deaths due to fentanyl sold illicitly on the streets. In Indiana, naloxone was administered by EMS units more than 20,000 times in 2021, and more than 2,800 Hoosiers died from a drug overdose in the 12 months ending in July 2022.
All of that means there’s a lot of room for improvement in how we address this complex issue. National efforts bring opportunity for improvement, but to truly effect change depends on local programs that directly prevent and treat opioid use disorder (OUD), prevent overdose death among persons using drugs, and support persons with opioid use disorder who are in recovery. Implementing effective programs requires a deep understanding of the myriad details that contribute to the problem.
OUD Requires a Multi-Agency Solution
For many states, addressing the opioid crisis requires collaboration among a variety of local partners involved in the epidemic—public health practitioners, emergency responders, treatment professionals, and law enforcement officials, among others. In addition, people who experience substance use disorder require social supports. Meeting basic needs like housing and food is essential for recovery.
To effectively serve people in need, states require detailed information that determines who is at greatest risk of overdosing, where people are overdosing, and where services are located. Without it, they can’t effectively design, target, and evaluate interventions. A 360-degree view of the problem provides the deeper understanding that enables effective solutions.
Interagency data sharing not only brings together the wide range of information required to determine where services are needed but utilizes that data to understand the effectiveness of programs. A premier challenge for public policymakers and public health practitioners is determining which initiatives will have the most impact, which measures are producing results, and where more resources need to be allocated.
Additionally, we need to use predictive analytics to move upstream and anticipate and prevent opioid use disorder and overdoses in the first place. Measures including developing educational initiatives for both schools and communities, determining effective treatment center placement, establishing recovery services, and identifying high-risk individuals are vital components of proactively combating the epidemic.
Probabilistic Record Linkage Brings Greater Possibility for OUD Programs
Public health practitioners need to examine the opioid crisis from both a population and an individual level so they can view the opioid crisis as a comprehensive ecosystem but also understand the risk pathways and care gaps experienced by individuals in a particular community.
Performing these analyses requires a longitudinal database that draws from a variety of source systems and links those data and events by person to provide a cohesive, comprehensive overview of any individual’s utilization of state programs.
The key to building that essential database is probabilistic record linkage, which uses artificial intelligence to overcome typos and mismatches and accurately links records by person. Linking records that do not include a personal identifier like a social security number has historically been a challenging process. Any difference across agencies in how data is collected (middle initial versus middle name, numerical versus written dates) or a data entry mishap will keep records from being linked by a deterministic algorithm, keeping states from making important connections, like how many of the same people are utilizing services to identify trends.
Even records with a robust identifier (e.g., SSN, driver’s license ID) won’t be matched without relying on human intervention in the case of transposed digits or other data entry errors. All source data—across all agencies—must be clean, complete, and accurate for deterministic record linkage to dependably match records. In the siloed public sector, that’s of course about as common an occurrence as getting struck by lightning during a blue moon. And therefore, capturing matches becomes a process as faulty as the data—and an inefficient one, at that.
- Probabilistic record matching presents a more accurate method that overcomes common situations such as:
- Data quality issues—typos, misspellings, missing or extra letters
- Data incompleteness—last four of SSN, year of birth, middle initial, nicknames
- PII mismatch, as when one database collects date of birth while another collects age
- Lifestyle changes—marriage, divorce, change of address
Probabilistic record matching considers the strength of a match across multiple data points. Which means records where, say, five out of six fields match and the last is a digit or letter away can be counted as matches but may have been missed by using deterministic matching. It’s incredibly effective, as well. Resultant’s probabilistic record linkage algorithm only has a 0.6% rate of false positives; overall, it’s 97% accurate. This gives states a significant boost in how they can evaluate and design programs to address OUD.
Achieving Better Outcomes Starts with Data Modernization
Public health programs must be nimble by definition. What challenges a community today lives only in history books a decade later, and who knows when the next pandemic will occur? The path toward a more effective response to the drug overdose crisis is one that can also bring tremendous benefit to every program in the agency.
That path is data modernization.
If most of your data systems are siloed and built using old technology, data modernization can sound daunting, particularly since more than half of IT projects fail. That nightmare statistic comes from some harsh realities that are easily overcome: failing to assess your current state and create a detailed, phased plan for proceeding—one that includes leadership engagement and other critical organizational change management details—is a sure way to doom a project.
Here’s another daunting reality: The need for modernization isn’t going away. Legacy systems will only continue to fall behind and prevent innovation. Without critical updates, the inter-agency data sharing that facilitates better outcomes for citizens just isn’t possible. With a modern infrastructure, analytical potential grows exponentially.
Getting there doesn’t have to be disruptive if you approach it wisely:
- Align leadership.
Ensure modernization is prioritized from the top and that the right change-leaders are in place to build momentum. - Determine desired business outcomes.
Identify the outcomes you want to achieve and the data you need to achieve them. - Understand your current state.
A thorough understanding of current-state operations, data, and infrastructure provides the foundation for your modernization. - Plan foundational infrastructure for core systems.
Design an infrastructure that utilizes modern technology and focuses on the interoperability, modularity, automation, and efficiency you need. - Create a data modernization roadmap.
Document and prioritize concrete actions to advance your data maturity, tying KPIs to your business goals. - Optimize data governance practices.
Develop or refine a data governance program to ensure your data is high-quality, well-managed, and secure. - Implement your roadmap.
Success comes from a phased approach that prioritizes your work and utilizes public-private partnerships. - Plan for sustainability.
Fund your data modernization pursuits for the long haul with strategies like braided funding and Federal Financial Participation.
The Real Risk Lies in Sticking with the Status Quo
In the face of an epidemic that is only growing and claiming more lives as it spreads through communities, the status quo isn’t a viable option. Unfortunately, despite heroic actions by those responding on the ground, we just have not seen the results we desire. Which means more bad health outcomes, more struggle, and more lives and families impacted.
As a next step, we need to take a cold, hard look at the situation on the ground. That means evaluating and modernizing your current systems in service of reaching the data maturity that provides a complete picture of the epidemic in your state. It’s a big job but will help you tackle the immediate crisis and prepare you to better serve your residents during the next public health emergency.
Learn more about our Health and Human Services offerings
Share:
About the Author
Kathy Lofy
Senior Consultant @ Resultant