Harold Rogers Update 2017 Melissa McPheeters, PhD, MPH
We Need a New Normal Full collaboration among diverse departments Appropriate integration of data from disparate data sources Business intelligence approach to rapid reporting and analysis Shared visibility and intelligence to identify inflections in the epidemic and opportunities to intervene together We need to treat this like the emergency it is
Collaboration of Efforts Joint collaboration between TDH, TBI, & TDMHSAS w/core group that meets regularly Identify potential data sources and how data would be used Establish key indicators to track (e.g. prescribing, overdoses, criminal activity) Development of dashboard to track changes/inflections or locate hot spots Rapid response effort to be employed
What have we been up to with data? Finding the data (not always easy) Negotiating which data we need and in what format (reports do not equal data) Integrating data into one system (entity management) Building access to the data (legal and technical possibly the hardest part) Building relationships (possibly the best part!) Working through plans for visualization
System Layout
The REAL system
Data Architecture (Enterprise Data Warehouse)
Data Use Pivot Tables Integrated data delivery into Excel directly from Warehouse tables Tableau Business Intelligence dashboards provide visual representations of data with continual or scheduled updates SAS or SAS Visual Analytics R Statistical Analyses can be run after pulling the data down into Desktop or Server versions of SAS R in SQL Server for predictive analytics Cubes SQL Server Analysis Services Cubes contain data pre-aggregated for improved performance of analytical queries
Grant Activities Regular meetings to understand what drives activity by our partners Discussions of what data could be made available Longer working sessions to design a dashboard Contact with legal teams to develop DUAs Short term, temporary solutions, including counties of concern index
Priority Indicators First priority indicators Overdose deaths, with breakdown by drug type (including buprenorphine, heroin and fentanyl) Drug overdose ED visits and hospitalizations Number/rate of opioid analgesics per 1000 residents Proportion of patients receiving greater than 90 MME of opioids Rate of multiple provider episodes (5/5/90) Proportion of patient Rx days with overlapping opioid and benzo prescriptions Number and rate of NAS births per 1000 live births by region MME for pain and MME for MAT and total MME Number of requests from healthcare providers in the CSMD Secondary indicators Number of registered users in the CSMD Number, rate and MME of buprenorphine prescriptions per 1000 residents Law enforcement and drug court requests to CSMD Combination of opioid, benzo and muscle relaxant Ratio of number of prescriptions to number of requests MME dispensed by top 50 prescribers Payer mix Opioid cough medicine use Naloxone saves and repeaters (no data available at this time)
Assisting with Investigations 2 tools developed: SAS tool provides lists and basic data based on complaint-driven information New tool in beta testing New, more flexible version in beta testing now built in house Includes linkage of CSMD data to mortality data Able to identify providers with pre-specified numbers of patients who have died, have certain prescription patterns or other characteristics
Data for Action Harold Rogers grant works side by side with the Prevention for States grant awarded by the CDC Hal Rogers grant provides access to drug arrest and felony conviction data, medical examiner case management system for fatal overdose deaths, and admission numbers to treatment facilities Syndromic surveillance will also be a component to track non-fatal overdoses
Data Dashboard-Test
Future objectives To add additional data, including: infectious disease data (i.e. HIV, HepC, endocarditis) & pain clinic data Integration and linkage of all new data into the PDMP Weekly tracking on data dashboard Team effort and agreement over review of data and how to act on it