October 10, 2018 •Laura Hoesly
Last week, Olivia Hebner and Laura Hoesly attended the Government Analytics Breakfast Forum: Using Analytics to Combat the Opioid Crisis sponsored by Johns Hopkins University and REI Systems. The speakers included Allison Oelschlaeger (Centers for Medicare and Medicaid Services), Dr. Mona Siddiqui (Health and Human Services), and Dr. Jim Kyung-Soo Liew (Johns Hopkins University). During the forum, speakers and audience members discussed data sources available to the federal government, challenges the government faces when analyzing the data, steps already taken, and possible methods for analyzing the data going forward.
What Public data are available?
The panelists discussed the wide array of publicly available datasets that are relevant to the opioid crisis, including 911 data from the Department of Transportation, opioid-related crime cases from the Department of Justice, prescription and treatment data from Centers for Medicare and Medicaid Services, or death information from the Center for Disease Control and Prevention. Lack of data is not an issue the government faces; Dr. Siddiqui noted that the government is, “Data rich but information poor,” meaning that there is plenty of data available, but not enough has been done to use the data to inform key decisions.
How can the Government better use their own data?
With this in mind, the panelists segued into the challenges that the government needs to overcome in order to become more information rich. Among the challenges are increasing data discoverability, streamlining data sharing while maintaining data privacy, improving data consistency across programs, reducing the lag time on data, improving data linking, and most importantly, focusing data efforts on the research questions the agencies want to answer.
CMS and HHS have already taken many steps toward resolving these issues. CMS developed a roadmap to address the opioid epidemic, with three pillars toward moving forward: (1) prevention, (2) treatment, and (3) data. HHS held a code-a-thon and conducted weeks of interviews and focus groups to develop a report summarizing the state of data sharing in HHS. CMS also provides the Research Data Assistance Center, which allows researchers to find, request, and use CMS data.
Using Data Analysis to Answer Policy questions and Improve Outcomes
Looking toward the future, Dr. Kyung-Soo Liew discussed some of the potential analysis techniques as the government’s data improves. For example, applying machine-learning techniques to build predictive models to estimate the probability a specific patient will become addicted based on their history, genome, and other observable factors. Other possible analyses that can help address the opioid crisis include analyzing alternative pain treatments, better understanding why certain patients did not become addicted, and developing alternative payment structures for doctors to treat patients.
The opioid crisis is not unlike most projects that involve data, where it is challenging yet critical to identify which datasets to use and how to transform the data into a usable format. Data scientists can help address various data issues such as reconciling different file formats, different data structures, missing data, and outliers. Once the data are prepared, statisticians and econometricians can apply cutting-edge proven techniques to the data to answer key policy questions and compile the results in a clear, understandable format to aid decision-makers.
Click here for an example of how Summit’s data scientists, statisticians, and econometricians have used the data that is already available to analyze costs associated with the opioid crisis.
Do you have a project that requires experienced data scientists who can help you overcome your data challenges? Contact us!
